MODELLING WITHIN-STORM

SOIL EROSION DYNAMICS

Contract ENV4-CT97-687

final report 1/4/1998 – 30/6/2001

Coordinator: CNR - Istituto per la Genesi e l’Ecologia del Suolo (CNR-IGES) Resp. scientist: Dr. Xxxx Xxxxx

Address: Istituto per la Genesi e l’Ecologia del Suolo

Xxxxxxxx xxxxx Xxxxxxx 00-00 - X-00000 Xxxxxxx, Xxxxxx

Telephone: x00 000 0000000

Fax: x00 000 000000

email: xxxxxxx@xxxx.xx.xxx.xx

Contractor: Katholieke Universiteit Leuven (KULEUVEN)

Resp. scientist: Prof. Dr. Xxxx Xxxxxx

Address: Laboratory for Experimental Geomorphology Xxxxxxxxxxxxxx 00 - X-0000 Xxxxxx, Xxxxxxx

Telephone: x00 00 00 00 00/26

Fax: x00 00 00 00 00

email: xxxx.xxxxxx@xxx.xxxxxxxx.xx.xx

Contractor: Cranfield University, Silsoe (CRANFIELD)

Resp. scientist: Xxxx. Xxx Xxxxxx

Address: Xxxxxx Xxxxxxx, Xxxxxx

Xxxxxxxxxxxx XX00 0XX - Xxxxxx Xxxxxxx

Telephone: x00 (0)0000 000000

Fax: x00 (0)0000 000000

e-mail: x.xxxxxx@xxxxxxxxx.xx.xx (Xxxx. Xxx Xxxxxx)

Contractor: Consejo Superior de Investigaciones Científicas (CSIC-CEBAS) Resp. scientist: Dr. Xxxx Xxxxxxxxxx

Address: Centro de Edafología y Biología Aplicada del Xxxxxx Xx. Xxxx. Xxxxxxxx 0000 - X00000 Xxxxxx. Xxxxxx

Telephone: x00-000-000000

Fax: x00-000-000000

email: xxxxx@xxxxxx.xxxxx.xxxx.xx

Contractor: Istituto di Idraulica Agraria (IIA)

Resp. scientist: Xxxx Xxxxx

Address: xxx Xxxxxxxxxxx, 0 - 00000 Xxxxxxx - Xxxxxx

Telephone : ++ 39 - 095 - 350401

Fax: ++ 39 - 095 - 0000000

email : xxxxx@xxxx-xx.xx.xxxxx.xx

Contractor: Institut für Bodenforschung, Universität für Bodenkultur (BOKU) Responsible Scientist: Dr. Xxxxx Xxxxxxx

Address: Xxxxxx-Xxxxxxxxxxxxx 00 X-0000 Xxxx, Xxxxxxx

Telephone: x00 0 0000 00000 00

Fax: +43 1 0000 00000 0

E-mail: xxxxx.xxxxxxx@xxx.xx

Contractor: Istituto Sperimentale per lo Studio e la Difesa del Suolo (ISSDS) Resp. scientist: Dr. Xxxxx Xxxxxxxx

Address: Xxxxxx X’Xxxxxxx, 00 - 00000 Xxxxxxx - Xxxxxx

Telephone: x00 00 0000000

Fax: x00 00 000000

email: xxxxxxxx.xxxxx@xxx.xx

Contractor: Faculty of Geographical Sciences, Utrecht University (UU) Resp. scientist: Dr. Xxxxxx Xxxxxx

Address: Xxxxxxxxxxxxxx 0

XXXXX 00000 - 0000 XX Xxxxxxx, Xxx Xxxxxxxxxxx

Telephone: x00 0000000

email: x.xxxxxx@xxxx.xx.x x

EU responsible: Xxxxx Xxxxx - DG Research, Brussels, Belgium

I Project Executive summary

II Partner’s Reports

• CNR - Istituto per la Genesi e l’Ecologia del Suolo (CNR-IGES)

• Katholieke Universiteit Leuven (KULEUVEN)

• Xxxxxxxxx University, Silsoe (CRANFIELD)

• Consejo Superior de Investigaciones Científicas (CSIC-CEBAS)

• Istituto di Idraulica Agraria (IIA)

• Institut für Bodenforschung, Universität für Bodenkultur (BOKU)

• Istituto Sperimentale per lo Studio e la Difesa del Suolo (ISSDS)

• Faculty of Geographical Sciences, Utrecht University (UU) III Technical Implementation plan (TIP)

I Project Executive summary

Abstract

Objectives

The projects aims were:

1) description of the within-storm dynamics of soil surface roughness, sealing, soil aggregates and infiltration

2) prediction of ephemeral gully generation and development during erosive storms

3) production of a generator of synthetic erosive rainstorms

4) development of a fully dynamic model taking into account point 1 and 2

5) development of a pedo-algorithm package for model users.

Results

All the goals have been addressed. Scientific achievements regarding 1, 2 and 3 have been obtained and now a description of interrill dynamics is available: surface roughness behaviour, sealing and infiltration dynamics have been investigated and original results obtained. They include the dynamics of ponded areas, saturated hydraulic conductivity and soil detachability (by splash). Gully generation is now clearly described in empirical and theoretical terms. A rainfall generator, suitable for this within-storm erosion modelling, has been produced. A set of pedo-algorithms has been developed, based on foreground knowledge and on original acquisition within this project. Two models (EUROSEM and EUROWISE) contain part of the developed knowledge. In particular they contain a routine for generating ephemeral gullies and an improved infiltration equation.

Introduction

The scope of the project was substantially that of producing one or more models describing soil erosion during extreme rainfall events allowing for surface modifications and for ephemeral gully generations (surface modifications explored in WP1, ephemeral gully in WP2). This led to the decision to modify two models, namely EUROSEM and LISEM, which are substantially based on the same type of mathematical representation of the physics of soil erosion but which differ in one extremely important aspect, i.e. The way they represent the landscape (the former being a conceptual representation - through a cascade of planes, the latter raster). Also different types of difficulty are encountered when generating ephemeral channels that were not already given as an input at the time of the preparation of the input tables and maps. WP4 was used for the coding of the new algorithms, and the validation of the new models.

When presenting a model, one of the most important elements is indeed the input data needed to make it run. Data collection is expensive, sometimes impossible, hence techniques helping experts to make good guesses at the input values are always needed. This brings in the problem of pedotransfer functions, of meteo data, etc. The part regarding a generator of synthetic rainfall was dealt with using WP3 where the model RAINGEN was further developed and improved. Better tables and algorithms for estimating appropriate soil characteristics were developed partly within WP4 and partly within WP1 and 2.

The results obtained by the project can be subdivided into 3 categories: 1) new knowledge developed into new routines in the two models; 2) new knowledge that was not possible to

incorporate into the models and 3) new methodologies.

These achievements have been publicised through scientific publications, many still only at the

submitted but not yet published stage, presentations at scientific conferences, presentations at disseminating conferences. Some material is already accessible as freeware downloadable from several web-sites, others will soon be so.

The following chapters are organised as follows: 1) summary of the activity ; 2) achievements grouped in the 3 said categories; 3) results and their diffusion ; 5) detailed summary of each partner's activity; 4) technical implementation plans.

Summary of project activities

Surface conditions are relevant for several processes affecting soil erosion, the main ones being surface roughness, which affects splash detachment and flow transport capacity, and sealing/crusting, which affects runoff generation and soil detachability.

Before discussing the various results obtained by the project, attention must be drawn to the effect of dispersion and of water quality in rainfall simulation (because most of the reported experiments were made in the laboratory with simulated rainfall). It was found (see CNR report) that tap water usually completely depresses dispersion differently from good quality water (electrical conductivity between 10-40 □S/cm) and rain water. This causes a slowing down of many processes linked to sealing and soil loss. In particular, soil loss can be 3 times greater when rain-water is used instead of tap water. This means that most of the interrill splash tables, equations or nomographs (usually based on rainfall simulation data) cannot be used and must be substituted by new ones.

Collected data have clearly indicated that Xxxxx and Xxxxxxxx (1978) equation usually fails when the soil has strong aggregates which slow down sealing processes (CNR report). Another equation has been tested (Xxxxx- Xxxxxxx and Xxxxxx, 1981) which usually compares favourably with Xxxxx and Xxxxxxxx. The intense dynamics of seed-bed surfaces suggest that none of these equations really adhere to reality. This is because the geometry of aggregates and particles changes fast during the first phases of clod destruction. Hence parameters taken as constant in the said equations must vary. When this is done it is possible to perfectly describe observed behaviours. Unfortunately, the equations describing the time-dependence of the hydraulic characteristics of the first layer is not analytical and many empirical equations can do the job equally well. This leads to an important uncertainty that was not solved within the project.

Surface roughness effects on maximum depression storage and on detention storage (see Cranfield report) were examined and data were collected using stereo photo and photogrammetric techniques. This led to some critique of the equations already suggested in EUROSEM manual for evaluation surface storage. Another set of experiments (CNR report) has led to the construction of a nomograph with simple and clear indication of the behaviour of roughness (relative standard deviation or relative random roughness index) in relation to rain impact energy and soil characteristics (permanent and transient). The nomograph describes well the results obtained in an independent set of experiments in the field with rainfall simulation (CEBAS report). Also the data collected in the field (under natural rainfall) can be easily located in the nomograph (ISSDS report). This result is not included explicitly within the models but will give indications for assigning initial roughness values in the tables of model input data.

The variation during the time that the area was effectively ponded (through series of images taken using a CCD camera connected to a computer) led to the identification of the relative importance of zones sealed by direct drop impact (erosion crust) and those sealed by deposition of splashed particles (CNR report). The results allowed for a completely different approach to soil detachability determination. Actually an inverse procedure for calculating soil detachability was devised. It is based on a splash detachability model proposed by Xxxxx et al. (1998), and takes into account the expansion of ponded areas and the destruction of roughness elements (or their generation in case of excessive soil loss). The application of this procedure to a series of laboratory experiments has led to a new set of erodibility values. As erodibility is also variable with time (or better, with rainfall characterised through its kinetic energy) the new erodibility table will also give indications concerning this.

Aggregate resistance is a characteristic which was explored in relation to several experiments (CEBAS and CNR reports). Results indicate that aggregate behaviour can be used to infer soil erodibility as well as soil propensity to sealing (and consequently to soil hydraulic properties). Nevertheless results did not allow the determination of any clear and reliable relationship between aggregate stability indices and the said characteristics. At present aggregate stability can be used only as an index that can help in identifying reasonable input values for the models.

Ephemeral gully generation has been studied extensively in the field (see KULeuven report mainly, IIA and ISSDS reports). Relationships identifying morphological threshold conditions for gully generation (0.1 m2 of gully cross section) and for gully fan formation have been studied at length. This has allowed the identification of areas in the landscape where gully erosion can take place.

Collection of data from the literature, some experimental data and data simulated using existing models has allowed the identification of a relationship linking total discharge to channel width, which is one of the basic parameters needed for calculating flow aggressiveness. All this has been reflected into one of the models (EUROWISE, which is a derivation of LISEM, see UU report).

A series of experiments conducted in the laboratory have clearly demonstrated that flow shear stress is the key parameter representing flow aggressiveness (KULeuven and CNR reports). This brings soil shear strength and soil resistance characteristics to the fore: their variation during the year and the differences in soil resistance between successive soil horizons can explain the differences between winter and summer gullies, between gullies entrenching different types of soil horizon (KULeuven report).

Other experiments conducted in the field (CEBAS, ISSDS and CNR reports) and in the laboratory (ISSDS and CNR reports) confirmed an original theoretical model (CNR report). The model was then used to verify the basis of the empirical relationships used within EUROWISE. This control gave a very positive outcome as shown in the CNR report. These experiments also pointed out that Xxxxxxx'x equation, which states that velocity depends on discharge, slope and roughness, does not work during channel erosion: only discharge seems to be related to velocity in agreement with previous findings of Xxxxxx (1992).

Regarding the problem of model development it was decided to expand LISEM into EUROWISE. The latter model is able to deal with gully generation. As already stated, it incorporates the empirical relationships already mentioned. The philosophy is simple: when there is a substantial flow in the area that is prone to gullying then a gully can be excavated. Hence the more suitable form of the threshold equation for gully initiation was chosen on the basis of its more or less correct identification of the prone area. Then the channel was allowed to increase following discharge.

Deepening of the channel was then adjusted using estimated soil detachment and distributing it over the already calculated width. The results were compared with gully data measured in the field (UU and KULeuven reports).

While it was possible to include and test this in EUROWISE it was not so in EUROSEM (Cranfield report) because it had to be first translated from FORTRAN77 into Delphi Pascal (which means the program was restructured). This effort has produced a much better version of the model but bugs (inevitable in these cases) prevented any validation of the gully algorithm.

Both the models include Xxxxx- Seytoux and Verdin infiltration equation. It appears that the water balance (EUROSEM, Cranfield Report) is better with this equation than with Xxxxx and Parlange's. An evaluation of how the models worked outside the range of conditions in which they have been created was also made (ISSDS report). Here one version of EUROSEM was tested with obvious difficulties due to the still untested Delphi code together with LISEM (it was impossible to test EUROWISE as it was under coding until the end of the project). It was found that the hydrologic part of both models does not cope well with clay rich soil in typical subhumid Mediterranean conditions.

This part of the project was preceded by a careful study concerning soil moisture (ISSDS report)

which made it possible to estimate conditions for presence or absence of cracks. With this at hand it is possible to state that both models fail largely in presence of cracks and to a lesser extent when the soil is close to field capacity. Failure when cracks are present is due to the fact that none of the models incorporate any routine for cracks and flow through macropores. The fact that cracks are open most of the time strongly limits the applicability of EUROSEM and LISEM to Mediterranean condition xx xxxx xxxx soils.

In order to make a synthetic rainfall generator (RAINGEN) capable of producing precipitation causing soil losses similar to the corresponding natural rains (BOKU report) it was decided to proceed by examining a series of alternatives (design storm; markov chain methodology based on wet day/dry day probabilities;Xxxxxxxx-Xxxxx-type models; scaling models).

The scaling model was the only one giving an acceptable correspondence between soil losses simulated with synthetic rain and those simulated with the corresponding natural rains. The basic assumption of the scaling approach is that actual rainfall intensity of storm events is a self-similar stochastic process which can be described with a scaling coefficient and a series of 3 parameters. To further complicate the model it was found that the scaling coefficient was usually time dependant. Hence, it was necessary to fit H using piecewise linear regressions. Rainfalls are generated supposing: 1) their duration to be Weibull distributed; 2) their total depths to be hyperexponentially distributed; 3) incremental depths to be Gamma distributed and 4) rainfall intensity within a storm to be a self-similar (simple scaling) process.

While exploring erosion behaviour using similar simulated rainfall but with different storm intensity patterns (BOKU report) it appeared that erosion varies with intensity pattern but in different ways for different soils. This indicated that the erodibility/detachability dynamics has an important role to play.

Independently from the Boku' study, another investigation (ISSDS report) showed that, in the short term – about 10 years, rain characteristics have evolved in Tuscany towards larger aggressiveness (drying up of climate in the short period with increasing rain kinetic energy, higher rain maximum intensity in 30 minutes and longer average lag between successive events). This indicates that models such as RAINGEN will soon need further refinements in order to cope with a shifting climate (in short periods).

The two erosion models and the rain generator have sufficiently developed interfaces which should cause little difficulty to skilled users. Each model contains simple tools for helping users in selecting input values of the various parameters.

A more sophisticated system of pedotransfer functions and algorithms has been organised in the form of a Java applet (Soil Equation Interface – SEI; CNR report). Presently it is completely operational with respect to the calculation of parameters such as soil water potential, saturated soil- matrix hydraulic conductivity and net capillary drive. Suggestions on how to evaluate soil interrill detachability are also given together with estimation of soil surface roughness (and its value at different times after seed bed preparation), and sealing intensity.

In order to make the applet useful where soil roughness estimation is concerned, a series of pictures of sites at different degrees of surface roughness was made while at each site roughness was measured.

This was done in two areas (Murcia and Tuscany, see ISSDS and CEBAS reports). The series of pictures is completed by a short discussion regarding roughness and a quick method for estimating degree of roughness decay, based on estimated soil resistance class (CNR report).

Achievements

•Models:

1.EUROSEM4win: erosion model (Delphi environment) 2.EUROWISE: erosion model (PCRaster environment) 3.RAINGEN 1: rainfall generator (C++, Fortran 90)

4.SEI: pedo-algorithms for estimating soil characteristics (JAVA applet)

•New parts implemented in models:

in erosion models:

1.Xxxxx-Xxxxxxx and Xxxxxx infiltration equation 2.Ephemeral gully generation

3.Ephemeral gully erosion

in RAINGEN 1

1.scaling model with time dependent scaling factor

in SEI

1.water retention curves 2.saturated hydraulic conductivity 0.xxx capillary drive

4.soil detachability (and its time variation) 5.roughness estimates and time dependant behaviour 6.gully wall/bed erodibility in wet and dry conditions 7.seal thickness

•Contribution to know-how

Major achievements are those related to ephemeral gully erosion. They include: 1) a series of relationships between discharge and channel width, bridging between the already known trends for rills and rivers; 2) a better understanding of the processes driving gully erosion; 3) a theoretical set of equations describing periods of high intensities, i.e. those relevant for shaping new channels; 4) certainty that Xxxxxxx'x equation is not valid during erosion peaks.

Regarding interrill processes, the dynamics of ponded areas within single-rainstorms and over several rainfalls has been studied and related to rainfall kinetic energy. This has allowed the construction of an interrill model which can be used for calculating soil detachability and its dynamics taking into account all the factors known to interfere with it. In particular, sealing thickness can be predicted. Even if this part has not yet been included into any of the models it may greatly improve data collection and analysis, when used as an inverse procedure. The project has also clearly shown that infiltration equations must be modified taking into account a variable hydraulic conductivity or, more generally, taking into account pore system variations. On this account, an existing model linking porosity variations to rainfall impacting energy has been confirmed and improved.

Other important achievements are those linked to rainfall. The best way of synthetically representing rainfall for erosion prediction is by means of a scaling model with a time dependent scaling exponent. Also the fact that in ten years it has been possible to appreciate a significative change in rain erosivity patterns points to an extremely important conclusion: short term scenario analysis related to soil erosion should be able to cope with local trends of climatic variations.

•New and improved methodologies

The following new methodologies and/or existing methodologies have been developed/improved and successfully tested :

• measuring saturated hydraulic conductivity in crusts;

• measuring depression storage using stereo photos;

• calculating soil detachment by inverse procedure;

• calculating saturated hydraulic conductivity in rainfall simulation tests using an inverse procedure (and assuming valid an infiltration equation)

• measuring soil susceptibility to rill and gully incision through field and laboratory experimental procedures

• synthesising rainfall characteristics for RAINGEN

• neural network analysis for predicting periods of maximum occurrence of extreme events.

Publicizing project results

List of papers (published, in press and submitted):

Xxxxxxxx X., (2001). “Capacità di invaso idrico superficiale: dinamica, misura e stima”. Riv. Di Irr. E Dren. 48(2):13-20

Xxxxxxxx L., M.P. Xxxxxxxx Xxxxxxx, M.S. Xxxxx, X. Torri. “La dinamica degli orizzonti di superficie nel bilancio idrologico del suolo”,

xxxx://xxx.xxxxxxx.xxxxxx-xxxxxxx.xx/xxxxxxxxxxxx/xxxxx /xxx_xxxxx/xxx_xxxx/xxxx.xxx#xxxxxxxxx .

Xxxxxxxx, X., Xxxxx, X., Xxxxxx, X., Xxxxxxx, X.X., 0000. Effects of water quality on infiltration, runoff and interrill erosion processes during simulated rainfall. Earth Surface Processes and Landforms, 26: 329-342.

XXXXX, X. XXXXXXX, X.X & XXXXX, X.X. 1999. Evaluation of the EUROSEM model for the Catsop watershed, The Netherlands. Catena 37, 507-519.

Xxxxxxxx- Xxxx , M., Xxxxxxxx, V, and Albaladejo, J. Hydrological and erosional response to natural rainfall in a degraded semi-arid area of Southeast Spain. Hydrological Processes, 15, 557- 571. 2001

Xxxxxxxx-Xxxx , X., Xxxxxx Xxxxx, P, Xxxxxxxxxx, J. and Xxxxxxxx, V. Influence of vegetal cover on sediment size particle distribution under natural rainfall in a semiarid environment.Catena 38, 175-190. 2000.

Xxxxxxxx-Xxxx, M.; Xxxxxx, R., Xxxxxxxx, V, and Xxxxxxxxxx, J. “Diseño experimental mediante lluvia xxxxxxxx xxxx xx xxxxxxx xx xxx xxxxxxx xx xx xxxxxxx del suelo durante la tormenta”. Cuaternario y Geomorfología (accepted Marzo 2001).

Xxxxxxxx-Xxxx, M.; Xxxxxxxx, V, x Xxxxxxxxxx, X. Xxxxxxxxxx of process-based soil erosion models: a field experimental design Land degradation and Development (in press)

Xxxxxxxx-Xxxx, M.; Xxxxxxxx, V, x Xxxxxxxxxx, J. “Relations between erosion processes and sediment particle size distribution in a semiarid area of SE of Spain” Geomorphology (in press)

Nachtergaele J, Xxxxxx J, Vandekerckhove L, Oostwoud Wijdenes D, Xxxx M. 2001. Testing the Ephemeral Gully Erosion Model (EGEM) for two Mediterranean environments. Earth Surface Processes and Landforms, 26 (1): 17-30.

Nachtergaele J., Xxxxxx X., 2000. EGEM, a potential prediction tool for soil losses by ephemeral gully erosion in the Belgian loess belt?. Xxxxxxxx, D., Schiettecatte, X. (Eds.), Proc. Erosion contact group-meeting, March 11th, 1999, International Center for Eremology, Ghent Univeristy, Belgium. I.C.E. Special report No. 2/2000: 55-60.

Nachtergaele J., Xxxxxx X., Xxxxxxx X., Xxxxxx I., Xxxxxxxxxx X., Xxxxxxxxxxxxxx X., Xxxxxx G., 2001. The value of a physically-based model versus an empirical approach in the prediction of ephemeral gully erosion for loess-derived soils. Accepted for publication in Geomorphology.

Nachtergaele, J., Xxxxxx, J., 1999. Assessment of soil losses by ephemeral gully erosion using high- altitude (stereo) aerial photographs. Earth Surface Processes and Landforms, 24: 693-706.

Nachtergaele, J., Xxxxxx, J., Xxxxxxxx Xxxxxxxx, D., Vandekerckhove, L., 2000. From ephemeral to permanent gully: the medium-term evolution of the Kinderveld gully. In: Xxxxxxxxxxx, G. (ed.), Historical and present-day soil erosion processes in central Belgium. Guide of the annual excursion of the Belgian soil science society, June, 14, 2000. Pedologie –Themata, 9: 60-65/81-87.

Nachtergaele, J., Xxxxxx, J., Xxxxxxxx Xxxxxxxx, D., Vandekerckhove, L., submitted. Medium-term evolution of a gully developed in a loess-derived soil. Geomorphology.

Nachtergaele, X., Xxxxxx, J., Xxxxxxxxx, X., Xxxxx, D., submitted. Flow width – discharge relations for rills and (ephemeral) gullies. Hydrological Processes.

Nachtergaele, J., Xxxxxx, J., Xxxxxxx, A., Xxxxxx, I., Xxxxxxxxxx, B., Xxxxxx, G., 2000. Ephemeral gully erosion in the Belgian loess belt. In: Xxxxxxxxxxx, G. (ed.), Historical and present-day soil erosion processes in central Belgium. Guide of the annual excursion of the Belgian soil science society, June, 14, 2000. Pedologie –Themata, 9: 56-59/77-80.

Nachtergaele, J., Xxxxxx, X., Xxxxxxx, A., Xxxxxx, I., Xxxxxxxxxx, X., Xxxxxxxxxxxxxx, L., Xxxxxx, X., 1999. Prediction of soil losses by ephemeral gully erosion using EGEM (ephemeral gully erosion model). Pedologie-Themata, 6: 76-85.

Nachtergaele, J., Xxxxxx, J., submitted. Spatial and temporal variations in resistance of loess- derived soils to ephemeral gully erosion. European Journal of Soil Science.

Nachtergaele, J., Xxxxxx, J., Vandekerckhove, X., Oostwoud Wijdenes, D., Xxxx, M., in press.

Testing the Ephemeral Gully Erosion Model (EGEM) in Mediterranean environments. Proc. 10th International Soil Conservation Organization (ISCO) Conference: Sustaining the Global Farm. Local Action for Land Stewardship. Purdue University, West Lafayette, Indiana, USA, 23-28 May, 1999.

Xxxxxx, X., de Xxxx, X., Xxxxxx, X., Xxxxxxxxxxxx, J., Xxxxxx, G., 1999. Concentrated flow erosion rates as affected by rock fragment cover and initial soil moisture content. Catena, 36: 315- 329.

Xxxxxx, X., Xxxxxxxxxxxx, J., Deckers, J., 2000. Gullies in the Tersaart forest (Huldenberg): climatic or antropogenic cause? In: Xxxxxxxxxxx, G. (ed.), Historical and present-day soil erosion processes in central Belgium. Guide of the annual excursion of the Belgian soil science society, June, 14, 2000. Pedologie –Themata, 9: 40-51.

Xxxxxxx, X.X. & Xxxxxxxxx, X. 1999. Modelling the impact of live barriers on soil erosion in the Pairumani sub-catchment, Bolivia. Mountain Research and Development 19: 292-299.

Regüés, D., Torri, D. Efecto de las Tormentas sobre la Dinámica Superficial del Suelo y su Relación Con La Formación de Cárcavas en Ambiente Mediterráneo. Cuaternario y Geomorfologia (submitted).

Xxxxx X. , J.; Xxxxxxxx-Xxxx, M.; Xxxxxxxxxx, J. and Xxxxxxxx, V. Organic carbon and nitrogen influenced by vegetation removal in a semiarid Mediterranean soil. Biogeochemistry (in press).

Xxxxxxxx S., X.Xxxxx, X.Xxxxxxx. Effect of rain on the macroporosity at the soil surface. European Journal of Soil Science (accepted).

Xxxxxxxxx, X., Xxxxxx, X., Xxxxxxxxxxxx, J., 1999. Xxxxxxxxxxxx xxx xxxx xxxxxx xxxxxxxxxxxxxxx xxxx xxxxxx xxxx xxxxxxxxxxxxxxx xxxxxx. Xx Xxxxxxxxxxxxxx, 0: 11-18

Xxxxxxx X. , A. Mentler: Phosphorus transport with sediment particles as affected by the location of rainstorm peak intensity. Proceedings of the OECD workshop on „Practical and innovative

measures for the control of agricultural phosphorus losses to water“. Xxxxxxxxxx Xxxxxxx xx Xxxxxxxxxxx xxx Xxxxxxxxxxxx, Xxxxxxxx Xxxxxxx, 0000.

Xxxxxxx X. , A. Mentler: Phosphorus transport with sediment particles as affected by the location of rainstorm peak intensity. Proceedings of the OECD workshop on „Practical and innovative measures for the control of agricultural phosphorus losses to water“. Xxxxxxxxxx Xxxxxxx xx Xxxxxxxxxxx xxx Xxxxxxxxxxxx, Xxxxxxxx Xxxxxxx, 0000.

Xxxxxxx X. , X. Xxxxxx, Xx. Xxxx, X. Xxxxxxx (accepted for TERRA): Diseño de un sistema de generación de lluvias para la Introducción de datos en el modelo EUROSEM. In press.

Xxxxxxx P., X. Xxxxxxx, X. Xxxx: Simulation zeitlich hochaufgelöster Niederschläge für hydrologische Modelle am Beispiel von EUROSEM (European Soil Erosion Model). Mitteilungen der Deutschen Bodenkundlichen Gesellschaft. Submitted

Xxxxxxx P., X. Xxxxxxx: Using a simple scaling model to generate synthetic rainfall input data for EUROSEM. In preparation.

Xxxxxxx X., X.Xxxxxxx, W.E.H. Xxxx: A rainfall generation procedure for the European Soil Erosion model (EUROSEM). Hydrology and Earth System Sciences, 3, 2, 213-222, 1999.

Xxxxxxx X., X.Xxxxx, X.Xxxxxxx, A. Mentler: Rainfall Simulation for Outdoor Experiments. In: X. Xxxxx, X. Xxxxxxx (eds.): Current research methods to assess the environmental fate of pesticides. pp. 329-333, INRA Editions, 2000.

Xxxxxxx X.: Raingen – A scaling model to generate event based rainfall. Concepts and technical description. Institut fuer Kulturtechnik und Bodenwasserhaushalt, Bundesamt fuer Wasserwirtschaft, Petzenkirchen 2001.

Xxxxx D. and Xxxxxx X..P.C., Modelling within-storm soil erosion dynamics. Presented at European Climate Science Conference, Vienna, 19-23 Oct. 1998, on CDROM.

Xxxxx X., X. Xxxxxxxx (2001). “Equations for high rate gully erosion”, Catena, (submitted)

Torri D., Xxxxxx Xxxxx D., Xxxxxxxxxx X. and Xxxxxxxx P. Within-Storm Soil Surface Dynamics and Erosive Effects of Rainstorms, Catena, 32, 8, 131-150.

Xxxxx, X. & X.X. Xxxxxxx 2000. Sensitivity analysis of EUROSEM using monte carlo simulation: I hydrologic, soil and vegetation parameters. Hydrological Processes 14 (5):915-926.

Xxxxx, X. Xxxxxxx, J.N. & Xxxxxx, J.A. 2000. Sensitivity Analysis of EUROSEM using Monte Carlo simulation: II the effects of rills and rock fragments. Hydrological Processes 927-939.

Xxxx X., X. Xxxxxxx, X. Xxxxxxx, W.E.H. Xxxx (Mitteilungen des hydrographischen Dienstes): Erzeugung von Niederschlagsereignissen für das Europäische Bodenerosionsmodell EUROSEM unter der Verwendung des Xxxxxxxx-Xxxxx-Xxxxxx-Xxxxxxx.Xx press.

paper presentations at conferences

Xxxxxxxx X., X. Xxxxx Xxxxxxxx Xxxxxxx, Xxxxx X. Xxxxx and D. Torri Dynamics and Properties of Ponding Areas International Symposium “The significance of Soil Surface charcateristic in soil erosion”. 20-22 sept. 2001. Strasbourg

Xxxxxxxx-Xxxx, M.; Xxxxxxxx, V. x Xxxxxxxxxx, J. “Change in soil and surface properties within- storm in a Mediterranean area of Southeast Spain”. Geographical perspectives on environmental management in Iberia Session. RGS-IBG Annual Conference. Plymouth. England. 2001. Oral Comunication.

Xxxxxxxx-Xxxx, M.; Xxxxxxxx, V. x Xxxxxxxxxx, J. “Modelling changes in soil erosion parameters during the storm” Third International Conference on Land Degradation Rio de Janeiro 2001. Oral Comunication.

Xxxxxxxx-Xxxx, M.; Xxxxxxxx, V. x Xxxxxxxxxx, J. “Within storm erosion dynamics in Mediterranean agricultural lands: an experimental design.” Third International Congress of The European Society for Soil Conservation. Valencia, España. 2000. Oral Communication

Nachtergaele, J., Xxxxxx, J., 1998. Ephemeral Gully Erosion Assessment for the last 50 Years via High Altitude Stereo Aerial Photographs. Case Study: The Belgian Loess Belt. ESSC- workshop: Long-term Effects of Land Use on Soil Erosion In a historical perspective, Müncheberg, Germany, Sept. 11-13, 1998.

Nachtergaele, J., Xxxxxx, J., 1999. Ravijnerosie en ravijnerosie-onderzoek in de Belgische Xxxxxxxxxx. Bijeenkomst contactgroep erosie, March 11, 1999. Universiteit Gent, centrum voor eremologie, Gent, België.

Nachtergaele, J., Xxxxxx, X., Xxxxxxxx Xxxxxxxx, D., Vandekerckhove, L. and Xxxx, M., 1999.

Testing and evaluating the Ephemeral Gully Erosion Model (EGEM) in Southern Europe (SE- Spain and SE-Portugal) and the loess belt (Belgium). Ephemeral gully erosion studies, possibilities of joint research. USDA-NRCS, National Sedimentation Laboratory - Oxford Mississippi, 23-25 August, 1999.

Xxxxxxxxxxxx, X., Xxxxxx, X., Xxxxxxxxx, X., Xxxxx, X., 0000. Flow width prediction for concentrated flow on agricultural fields. Cost 623: ‘Snowmelt erosion and related problems’. The Norwegian State Pollution Control Authority (Jordfrosk), Xxxx, Xxxxxx, 00-00 Xxxxx, 0000.

Nachtergaele, J., Xxxxxx, J., Xxxxxxx, A., Xxxxxx, I., Xxxxxxxxxx, X. and Xxxxxx, G., 1999.

Ephemeral gully erosion in the Belgian Loess Belt. 2nd Int. Symposium on Tillage Erosion and Tillage Translocation. K.U. Xxxxxx, Xxxxxxx, 00-00 April 1999.

Nachtergaele, J., Xxxxxx, J., Xxxxxxx, A., Xxxxxx, I., Xxxxxxxxxx, L., Xxxxxx, X., 2000. Ephemeral gully erosion in the Belgian loess belt. International symposium on Gully erosion under Global Change, K.U. Leuven, 16-19 April, 2000.

Nachtergaele, J., Xxxxxx, X., Xxxxxxx, A., Xxxxxx, I., Xxxxxxxxxx, X., Xxxxxxxxxxxxxx, L., Xxxxxx, X., 1998. Prediction of soil losses by ephemeral gully erosion using EGEM (ephemeral gully erosion model). Gemeenschappelijke studiedag van de Belgische verenigingen voor bodemkunde en landelijk genie “Studie van bodem en duurzame ontwikkeling”. Centre de Recherches Agronomiques, Gembloux, Belgium, November 25, 1998.

Nachtergaele, J., Xxxxxx, X., Vandekerckhove, X., Oostwoud Wijdenes, D., Xxxx, M., 1999. Testing and evaluating the ephemeral gully erosion model (EGEM) in Mediterranean environments. 10th International Soil Conservation Organization (ISCO) Conference: Sustaining the Global Farm. Local Action for Land Stewardship. Purdue University, West Lafayette, Indiana, USA, May 23-28, 1999.

Xxxxxx, X., Xxxxxxxxxxxx, X., Vandekerckhove, X., Oostwoud Wijdenes, D., Xxxx, M., 1999. Prediction of ephemeral gully erosion in Mediterranean environments. I.A.G. Regional Conference on Geomorphology. University of Rio de Janeiro, Brazil, July 17-22, 1999.

Xxxxxx, X., Xxxxxxxxxxxx, J., Vandekerckhove, X., Xxxxxxxx-Xxxxxxxx, D., 1999. Datasets needed for predicting ephemeral gully erosion under global change. BGRG Rainfall Simulation Working Group Concluding Meeting (Joint Meeting with COST 623 Soil Erosion and Global Change). 00-00 Xxxxx 0000, Xxxxxxxxx, X.X.

Xxxxxx, X., Xxxxxxxxxxxx, J., Xxxxxxxxxxx, G., Oostwoud Wijdenes D., Xxxxxxxx, C., 2000. Gully erosion under environmental change. International symposium on Gully erosion under Global Change, K.U. Leuven, 16-19 April, 2000.

Xxxxxx, X., Xxxxxxxxxxxx, X., Xxxxxxxxxxx, X., Xxxxxxxxxxxxxx, X., Xxxxxxx, X., 0000. Gully erosion as a missing link in erosion models. COST 623 International workshop on Linkage of Hillslope Erosion to Sediment Transport and Storage in river and Floodplain Systems.

Almeria, Spain, 7-11 September, 2000.

Xxxxx X., J.; Xxxxxxxx-Xxxx, M.; Xxxxxxxx, X. and Albaladejo, J. The effect of vegetation removal

on soil organic carbon losses: a 9 years experiment in semiarid SE Spain. Third International Conference on Land Degradation Rio de Janeiro 2001. Poster.

Torri D. , X. Xxxxxxxx, X. Xxxxxxxxxx, M.S. Xxxxx, X.X. Xxxxxxxx Xxxxxxx, X. Xxxxxxxx, X.Xxxxxxxxx, X. Xxxxxxxxxxxx, X Xxxxxx, X Xxxxxxxxxx, X. Xxxxxxxx, X. Xxxx Experiments and Algorithms for Linear Erosion and their Evolution. International Conference: “Sustainable Soil Management for Environmental Protection: Soil Physical Aspects” Florence, Italy, 2-7 July 2001, Book of Abstracts, page 97.

Xxxxx X., X. Xxxxxxxx, X.Xxxxx Xxxxxxxx Xxxxxxx, Xxxxx X. Xxxxx Xxxxxx-Induced Soil Surface Dynamicsn International Symposium “The significance of Soil Surface charcateristic in soil erosion”. 20-22 sept. 2001. Strasbourg.

Web Sites

Some of the material has already been made available to users. Models are at the moment either downloadable from the web (freeware) or consultable on the web:

EUROSEM 4 win:

• xxxx://xxx.xxxxxx.xxxxxxxxx.xx.xx/xxx/xxxxxxx/xxxxxxx/xxxxxxx_xxxxxxxxxxxx.xxx

EUROWISE:

• xxxx://xxx.xxxx.xx.xx/xxxxx

SEI:

• xxxx://xxx.xxxx.xx.xxx.xx/xxxx/xxxxxx/Xxxxx_xxxxxx.xxx

• xxxx://xxx.xxxx.xx.xxx.xx/xxxx/xxxxxxxx.xxx

RAINGEN 1:

• xxxx://xxx.xxxx.xx.xxx.xx/xxxx/xxxxxxx/xxxxxxx.xxx

Distribution of results

Four technical implementation plans have been written, one per software packge. Explanations and development programmes are given there.

Other results such as methodologies are or will be described in scientific papers.

Information on other particular topics developed in the project can be asked directly to MWISED partners as listed below:

Infiltration, Ksat	CNR-IGES
Surface roughness, depression storage, ponding	CNR-IGES, ISSDS, CEBAS, CRANFIELD
Ephemeral gully	KULeuven
Rainfall characteristics	BOKU, ISSDS
Validation data sets for models at field and catchment scale	ISSDS, IIA, CEBAS
Soil erosion models	UU, CRANFIELD
Soil algorithms and pedofunctions	CNR-IGES

The Project Executive Summary can be found on the web at the following URL:

xxxx://xxx.xxxx.xx.xxx.xx/xxxxxx

II Partner’s Reports

• CNR - Istituto per la Genesi e l’Ecologia del Suolo (CNR-IGES)

• Katholieke Universiteit Leuven (KULEUVEN)

• Xxxxxxxxx University, Silsoe (CRANFIELD)

• Consejo Superior de Investigaciones Científicas (CSIC-CEBAS)

• Istituto di Idraulica Agraria (IIA)

• Institut für Bodenforschung, Universität für Bodenkultur (BOKU)

• Istituto Sperimentale per lo Studio e la Difesa del Suolo (ISSDS)

• Faculty of Geographical Sciences, Utrecht University (UU)

CONSIGLIO NAZIONALE DELLE RICERCHE ISTITUTO PER LA GENESI E L’ECOLOGIA DEL SUOLO (IGES)

REPORTING PERIOD: 1 APRIL 1998– 30 JUNE 2001

Contractor:

Responsible scientist:

Scientific staff: Address:

Telephone: Fax:

email:

Consiglio Nazionale delle Ricerche – Istituto per la Genesi e l’Ecologia del Suolo

Dr. Xxxx Xxxxx

X. Xxxxxxxx. M.P. Xxxxxxxx Xxxxxxx, X. Xxxxxxxx, L. D’Acqui, X. Xxxxxxxxx, X. Xxxxxx, M.S. Xxxxx.

Istituto per la Genesi e l’Ecologia del Suolo Xxxxxxxx xxxxx Xxxxxxx 00-00

X-00000 Xxxxxxx, Xxxxx

x00 000 0000000

x00 000 000000

xxxxxxx@xxxx.xx.xxx.xx

WORK CARRIED OUT DURING THE REPORTING PERIOD

Introduction

The IGES group has been involved in experimenting and in producing algorithms for soil surface dynamics, infiltration dynamics and gully generation.

This has brought to the definition of two original methodologies regarding rainfall and runoff simulations. Presently algorithms for random roughness decay and water saturated conductivity (Ksat) evolution have been prepared. Regarding ephemeral gully generation a set of theoretical equations have been developed in order to have a scientific background for designing experiments and drawing conclusions. The equations have been coded into a provisional model (GULLYER) in order to appreciate the several feedbacks.

WP1-ST1 – rainfall simulation in laboratory Methodology and results

The methodology that was set up consists of a series of rainfall simulations conducted

over the same soil sample (usually 4-5 rainfalls) (Borselli et al. 2001). Each set of experiment is here referred to as ‘history’. Histories are different because of rainfall characteristics (intensity, energy, and rainfall duration), initial soil moisture content, slope gradient. Soil is always prepared in order to reproduce a fine seedbed (the most risky situation for erosion, and the most dynamic because the surface undergoes the greatest modifications). Each rainfall simulation is complemented by soil roughness measurements taken using a laser profile-meter (Borselli, 1999). During later experiments the evolution of the surface of the ponds was followed by means of images taken with a CCD camera connected to a computer (10-15 images per rainfall). The water used for rainfall simulation is of good quality (conductivity smaller than 20 μm cm-1) because the strong influence on experimental results when tap water is used instead (Borselli et al. 2001).

Soil types

Laboratory experiments on samples collected in field were run in order to gain insight in the processes of soil surface modification under air-dried or field capacity initial moisture content. Soil samples were collected in different areas of Italy. Moreover soil samples (900 kg, ca.) from Belgian Loess were received from KUL Leuven (B) to perform experiments on them.

Table with main properties of examined soil

Soil code	Soil Classification U. S. D. A.	Lithology	Clay %	Silt %	S and %	O. M. %	pH 1:2. 5	Mineralogy
ORCIA 31	Chromic Calcixerert Typic Xerorthent Typic Eutrochrept Vertic Xerorthemt Xxxxxx Luvisol Ranker Sulfic Endoaquepts Aquic Ustochrept Udertic Ustochrepts	Pliocene Clays Pliocene Clays Olocene depos. Pliocene Clays Loess M ethamorphic rocks Deltaic deposits Alluvial deposits Alluvial deposits	53.85 40.70 5.45 53.40 45.50 1.10 23.00 42.70 34.30 42.00 43.00 15.00 10.00 78.00 12.00 8.00 43.80 48.20 16.00 40.00 44.00 17.00 66.00 17.00 35.00 57.00 8.00			2.50 2.20 1.60 0.50 0.42 7.04 9.90 1.20 2.40	7.40 8.30 7.90 8.20 7.80 4.69 7.60 7.90 7.40	Vermiculite, Clorite, Kaolinite, Interst. M i-Sm.
ORCIA 30								Clorite,Smectite,Illite, Q ,F,C
FAGNA								M ica,Vermiculite,I-V,Illite,Kaolinite,InterstCl-Sm
VICARELLO								M ica,Vermiculite,Illite,Clorite,Kaolinite,InterstM i-Sm.
LOVANIO								M ica, Xxxxxxxxx, Interst. V-Sm (disordinato)
SARDO
SINORG								Smectite, Clorite, M ica, Interst. Cl-Sm.
SI NMCB
SINRNV

Measurements of the Saturated Conductivity of topsoil sealing crust

The evolution of saturated conductivity (Ks) during the progress of each history was examined using a constant head permeameter and samples of surface crust collected from the plots at different stages of the rainfall test/histories. The surface crust (irregular form in section view) have generally a depth variable from 0.5 to 1 cm. a special technique allow to insert the sample inside a standard steel cylinder and seal the contacts with steel using silicon and complete the filling of the cylinder using coarse quartz sand (fig. IGES_1).

Two type of crust were examined: 1) the deposition crust made by accumulation of splash detached particles and interrill-overland-flow transported particles in local depressions and 2) the erosion crust formed in non submerged areas which were continuously subject to drop impact.

For some soil a complete evolution of Ks was drawn (Fig.IGES_2). Generally we observed a very rapid decay from the initial Ks values (typical of soil matrix of the Ap horizon far from the soil surface, hence not exposed to the direct effect of rainfall.

H20 flux

Crust

Silicon sealing

Sand

Steel cilinder

H20 flux

Fig. IGES_1: Sample preparation for the measurement of saturated conductivity of surface sealing crust

SINA MCB

SINA P3 RNV

Evolution of Seal-Crust Saturated Conductivity

(Direct measurements by constant head permeameter)

Seal-Crust Saturated Conductivity

KSC (mm h )

-1

0.1

0.01

0 500 1000 1500 2000 2500 3000 3500

Cumulative kinetic Energy (J m-2)

Fig. IGES_2: sample graph on the evolution of the saturated conductivity of surface sealing crust of two studied soils

Porosity

The original planned activity was the collection, at the end of each rainfall simulations, of samples for porosity measurements. This part has undergone a series of modification along the way because of accidents and because of initially poor results. Two lines were initially investigated: 1) identification of the pore size at which crusts are perforated through all its thickness, and 2) 3D reconstruction of the pore system to identify (applying techniques such as ‘percolation’) parameters relevant to the hydraulic behaviour of the soil.

The first approach was deluding because pore size resulted to be within 8 and 12 micron with no apparent dynamics. The second approach was extremely time consuming and the resolution imposed be the machinery (38 micron) resulted to coarse for appreciating real dynamics. Hence it was decided to turn to a more classical approach, through the examination of pore size distribution (from 30 micron up) in samples collected at different times during a succession of rainfall simulations. The results (accepted for publications in the European Journal of Soil Science) are shown in the graph below (Fig. IGES_3). The first graph refers to the decrease of one type of pores (the elongated one, which is the only type exhibiting any dynamics). The equations ruling the process are those by Xxxxxx at al (1997) modified for accounting the fact that one parameter, originally assumed to be constant, is actually varying. The second graph describes the evolution of the exponent (the fractal dimension) of the pore volume. Higher values indicate better spatially distributed pores. This also means that pores are much smaller and more disconnected, hence water fluxes are reduced.

The leading equations are:

a [P + P {1 − exp( −k ' E)}− P ]

∆P = a EL,0 EL,1 E ES k

E + (1 − a )P

{1 − exp( −k E}

ä S ES0 E

which represents the reduction of elongated porosity which is driven by rainfall kinetic energy (E), soil detachability (ks) and other parameters reflecting initial pore system characteristics; and

∆DV = a E + b{1 – exp (-c E)} in the upper 0-3 cm depth and

∆DV = b{1 – exp (-c E)} in the lower 3-6 cm depth where

N>d ∝ d (1− DV ) ,

where N > d is the smallest number of pixels with size d that cover the area occupied by pores (i.e.. it describes pore size distribution).

Decrease of elongated porosity

0.5

Increase of volume fractal dimension

0.4

10 0.3

Xxxxxx et al.,1997 these experiments

0.2

0 500 1000 1500

Impact energy / J m-2

0.1

0 0.5 1.0 1.5

-2

Raindrop impact energy /kJ m

Fig.IGES_3: Decrease of elongated porosity – lines are calculated using a modified Panini et al. (1997) model; Dynamics of the fractal dimension of the pore system while xxxxx develops.

Infiltration models

The processing of rainfall simulation data clearly showed the inadequacy of Xxxxx and Xxxxxxxx (SP) infiltration equation (Xxxxx & Xxxxxxxx 1978) used in the previous EUROSEM model (Xxxxxx et al. 1998). In order to make it able to describe the observed behaviour all the variance was attributed by final infiltration rate (variation of the effective Ksat parallel to SP infiltration equation) which is unrealistic. When the soil is dry sorptivity is important and cannot simply disappear. Xxxxx- Xxxxxxx (MS) equation (Xxxxx- Xxxxxxx & Xxxxxx, 1981) seems to perform much better and the MS infiltration curve is not the same as the Ksat curve. It suggest the substitution of the SP equation. Practically, the SP equation posticipates time-to- ponding and underestimate final infiltration rate or, viceversa anticipate time-to-ponding and overestimate final infiltration rate. In the former case runoff is strongly overpredicted while in the latter it is underpredicted. This is due to the structure of the equation which makes it ‘rigid’ (time-to-ponding and final infiltration rate being too strictly related) and always simulating fast infiltration rate decays. Examples are given in FIG.IGES_4, relative to some experiments conducted during this project. The same techniques were applied to data collected during a previous project in which EUROSEM was developed. (FIG.IGES_5)

Infiltration is clearly well represented when saturated hydraulic conductivity (and connected parameters, such as net capillary drive) are allowed to vary (Fig.IGES_8).

ORCIA31 soil - 6° storm

Static models fitting

t. ponding

rainfall rate

Measured i(t)

Ks(t) [MS & SP]

Inf. rate - Sat. Cond (mm/h)

0 10 20 30 40 50 60 70

Time (min)

ORCIA31 soil t - 6° storm

Dynamic models fitting

t. ponding

rainfall rate

Measured i(t)

DMS

Ks(t) [DMS]

DSP

Ks(t) [DSP]

Inf. rate - Sat. Cond. - (mm/h)

0 10 20 30 40 50 60 70

Time (min)

FIG. IGES_4: Non Linear fitting (Borselli, 1998) of Xxxxx and Xxxxxxxx (1978) and Xxxxx-Xxxxxxx & Xxxxxx (1981) equations using both static and dynamic approaches for two soils.

Static models fitting

rainfall rate

immediate ponding

8 0

Fagna soil/ EUROSEM 93 - Potatoes - test 3

Inf. rate - Sat. Cond (mm/h)

Measured i(t)

SP unable to fit

Ks(t) [MS]

6 0

4 0

2 0

0 1 0 2 0 3 0 4 0 5 0 6 0 7 0

Time (min)

Fagna soil/ EUROSEM 93 - Potatoes - test 2

Dynamic models fitting

i(t) DMS

Ks(t) DMS

i(t) DSP

Ks(t) DSP

Inf. rate - Sat. Cond. - (mm/h)

0 10 20 30 40 50 60 70

Time (min)

FIG. IGES_5: The same techniques shown in FIG.IGES_4 were applied to data collected in previous experiments.

The Xxxxx-Xxxxxxx & Xxxxxx (1981) equation as infiltration routine

The Xxxxx- Xxxxxxx & Xxxxxx (1981) infiltration model has been adopted as basic equation in an inversion procedure to derive the main soil hydraulic parameters of the studied soils; and as additional infiltration model in the both EUROSEM and EUROWISE codes.

The Basic Soil parameters considered in the model are:

G = net capillary drive or wetting front suction as in Mein-Xxxxxx /Green- Ampt models (in mm)

Ks = saturated conductivity (in mm/h)

A composite parameter B is defined as:

s i

B = G(q − q )

(as in Xxxxx- Xxxxxxxx model (1978)) where:

[1]

(q − q )

is the initial water deficit with respect the natural saturation

The model applies once time to ponding (defined as the instant at which the rainfall intensity is equal to the infiltration capacity in soil) is reached.

If r(t) is the ietograph (in mm/h) then ponding at time (t) occur when the following equation is verified (Xxxx & Xxxxxx,1973, Xxxxx- Xxxxxxx & Xxxxxx 1981):

tp =

Ks B r(t)2 − K

r(t)

[2]

in this case t = tp= time to ponding (in hours).

for a constant rainfall of intensity r (rainfall simulation experiments):

t p =

Ks B

r 2 − K r

[3]

for a variable rainfall equation (3) the following equation must be verified at each time step

t p =

Ks B

_ _

[4]

r 2 − K r

where r is the average rainfall since rainfall start until ponding occur:

_ 1 t p

r = ∫0

r(t )dt

[5]

Eq.(2) shows some difference with the time to ponding calculated in the previous version of EUROSEM (which used only Xxxxx & Xxxxxxxx, 1978) where:

G∆q  r 

t p =

ln  

r  r − Ks 

[6]

In the in the following graphs the functions (2,6) are compared with the help of two auxiliary and composite parameters:

a = G(qs −qi )

(in h) [7]

b = r

(adimensional) [8]

α = G(θ -θ)/r

s i

β = r/K

t =α 1/( β-1) MS

α = 3

t =α ln(β/( β-1)) SP

α = 1

tp(h)

0.1

0.01

1 10

100

Fig.IGES_6. comparisons between time to ponding as calculated and Xxxxx- Xxxxxxx & Xxxxxx(1981) in eq.(2a), by Xxxxx & Xxxxxxxx (1978) equation (SP) in eq.(2e).

Infiltration after ponding

if r (t ) > k s

2Ks (B + I p )

then





 1 

i(t) =

 t − t p

+ A  + Ks

[9]

where A is a composite parameter given by:

(B + I )2

A = p

 r(t)  2

[10]

2Ks B

 Ks

−1



and

I p =

t p rdt

∫

[11]

is the cumulative infiltration until the ponding time.

Dynamic form of the eq. (9)

The dynamic behaviour of MS equation is introduced by a time variable Ks that represent the evolution of a sealing surface by the formation of sedimentary crusts and porosity reduction during a storm over the hydrologic history of the soil.

The evolution of the

Ks(t)

follows a sigmoidal-shape decay curve as theoretically and

experimentally derived by some authors (Xxxxxx & Assouline 1990). In our case it is represented by a four parameters hyperbolic tangent function:

Ks = Ks

where:

0−(Ks0

− Ks f

) tanh(kta)

[17]

Ks 0 is the initial value of Ksat

Ks f

the final value of Ksat

k, a , best fitting coefficients

Aggregate stability

A technique for measuring aggregate characteristics, which was first defined during a previous EU project, then partly refined during an Italian project, has been re-examined and the methodology completely defined. It has been used for measuring the stability of aggregates for the soil used in the laboratory tests and some other experiments have been made for characterising other two soils whose infiltration data will be used for these project. A series of decay curves are shown in FIG.IGES_7.

The data show that the decay constant is related to several soil parameters such as sealing formation rate and hydraulic conductivity decay. Presently no clear trend has been isolated yet, hence the use of this parameter is still under judgement.

P1 - AE P1 - Bt P2 - A

6000

5000

4000

D 50 (μ m)

3000

2000

1000

0 10000 20000 30000 40000

Energy (J m -2 )

FIG.IGES_7: decay of the aggregate median size with applied energy for different soils and soil horizons.

Effects of water quality

The effect of water quality on erosion and runoff generation has been explored for a series of reasons: 1) the importance of dispersion and of the weakening of soil particle bonds is usually ignored 2) if dispersion is important than soil parameter values measured with rainfall simulation techniques based on tap water must be abandoned.

The study was conducted comparing soil behaviour on two soil types with tap and good quality water. The effects are shown in FIG.IGES_8 and are self commenting. It seems that erosion is 3 times larger when good quality water is used. This means that most of the input data suggested for models represent underestimation, which can be counteracted by the usually poor sedimentation routines.

Loess soil (Leuven, B) Tap water r=54 mm h -1

Infiltration rate (mm h -1 )

0 5 10 15 20 25 30 35 40 45 50

Time (min)

Loess soil (Leuven, B) Distilled water r=54 mm h-1

1st run 2nd run 3th run

4th run

1st run 2nd run 3th run

4th run

Infiltration rate (mm h -1)

0 5 10 15 20 25 30 35 40 45 50

Time (min)

FIG.IGES_8: effect of the chemistry of artificial rainfall on infiltration: time to ponding and final infiltration rates are reduced when good quality water is used.

Dynamics Modelling of Surface Processes

In order to give a unitary view to all the data collected a model was devised for interrrill erosion. It was that developed into a software for extracting erodibility values from experiments. It includes detachment in microheights, deposition and erosion in microlows, effects of local slope gradient, expansion of ponded areas (water cushion effect on detachment), decrease of roughness (an consequently of local slope gradient value).

The aim is that of using detachment rate as a best fitting parameter in order to obtain the correct final roughness and measured net sediment export rate graph.

The variations of soil detachability will be used to suggest mean erodibility values varying following storm duration and soil surface conditions (FIG.IGES_9,10,11).

SINA LCO

Storms on dry soil

wash erosion rate

1° 2°

3°

4°

0.020

wash erosion rate (Kg/min/m2)

0.015

0.010

0.005

0.000

0 500 1000 1500 2000 2500

-2

Cumulative kinetic energy (J m )

FigIGES_9: soil loss in a series of 4 successive rainfall experiments.

SINA LCO

Storms on dry soil

Evolution of Random Roughness

1°

2°

3°

4°

Random Roughness - RR (mm)

byRRFIT (0.2) code MWISED Project

0 500 1000 1500 2000 2500

Cumulative kinetic energy (J m -2)

Fig.IGES_10: The line represents the calculated variation of random roughness during the 4 rainfalls (Fig. CNR.2). The circles are the measured RR values.

0.35

splash (detach./transport) rate (Kg min-1 m-2 )

0.30

0.25

0.20

SINA LCO

Storms on dry soil

Mean splash detach. rate Net splash transport rate

Splash Detach. rate at 0% slope

storm 1 0.00103

storm 2 2.939E-4

storm 3 1.493E-4

storm 4 9.61E-5

0.15

0.10

0.05

0.00

1° 2° 3° 4°

byRRFIT (0.2) code MWISED Project

0 500 1000 1500 2000 2500

-2

Cumulative kinetic energy (J m )

Fig.IGES.11: calculated soil detachability

FIGG.IGES_12 and 13 show the dynamics of surface sealing. It has effect on Ksat and on other surface characteristics that influences soil erosion (e.g. Xxxxxxx’x n).

Fig.IGES_12: series of pictures of sealed surface as it expands while roughness is flattened

M CB:

air dry

field capacity

envelop curve

100

Sealed surface (%)

0 500 1000 1500 2000 2500 3000

Kinetic Energy (J m )

Fig.IGES_13: variation of sealed surfaces during successive rainfall on the same soil surface.

Roughness evolution - MWISED Project - CNR-IGES

0.8

RR0.6 R

0.4

Sardo1

Sardo2(stones)

Sardo3

Sardo4 SinOrgSecco SinOrg.Umido SinMCB

Sina P3RNV Orcia 30

Fagna Patate

0.2

0 1000 2000

3000 4000 5000

Cum. Kin. Energy (J/m^2)

6000

VicarelloStns00 Vicarello1-99 Loess

Orcia 31

7000

Fig.IGES_14: Random roughness decay for various soils and initial moisture conditions

Random Roughness was examined for 14 soils at different moisture initial conditions.

The results showed in Fig. IGES_14 indicate the existence of a set of conditions as exemplified in Fig. IGES_14

Relative RR decreases with speeds, which belong to 3 main domains. The more resistant group includes skeletal soil. The intermediate group includes soils on which rain fell when they where fairly wet. The third group contains soils unstable or initially air-dried.

Soil stability and

roughness evolution

model: RR/RR0 = (1-d) e +d

-bE

high erosion rate or skeletal soils

soil with high stability

d=0.5 b=0.0005

Antecedent moisture effects

soil with medium stability

d=0.2

b=0.001

soil with low stability

1.2

Relative random roughness (RR/RR 0)

1.0

0.8

0.6

0.4

0.2

0.0

0 1000 2000 3000 4000 5000

MWISED Project

CNR-IGES

Cumulative Kinetic energy (E) (J m -2)

Fig.IGES_14: Suggested subdivision of RR decay in classes

WP2

Gully Generation model

An original model for rill and gully evolution was developed during the second year of activity. The basic equations on which it is based are:

∂W =

∂t

2kS (e

p − pcr ,s )

(1)

∂D = 1 k

(p − p

)− D ∂W

− s U ⇒

∂t r B

⇒ ∂D = 1

∂t r B

cr ,b

(p − p

)−

cr ,b

∂t X

0 (x

x X x

sed

p x xxx ,x

)X − s U

sed

(2)

where D is channel depth, W is channel width, kS and kB are erodibilities of the side-walls and of the bed, p is flow shear stress pcr,s and pcr,b are threshold shear stress values needed to detach every type of grain or aggregate respectively from the walls and from the bed, ef is the efficiency of p to erode the wall (generally if 1 is the efficiency for bed/thalweg erosion then the efficiency on the wall is smaller), sL is sediment load and Used is sedimentation velocity in turbulent flow.

The terms kB and pcr,b are not proper constants: they vary following the time-dependent fraction of the bed that is covered by sediment deposited from the water, the fraction covered by wall sediment, and the fraction of bed where bed material outcrops.

The two equations cannot be solved explicitly for the general situation. Only solutions valid in particular cases can be analytically derived. Some are shown in the enclosed manuscript (presented at the Gully symposium in Leuven and presently submitted to Catena). The most interesting of these cases is when the channel section is completely inside the upper uniform soil horizon. If erosion is very intense (practically no deposition in the channel - The few experiments with deposition were removed from the pool) then width must grow proportionally to channel depth. This is what happens in all the experiments.

For this special cases D and W must grow proportionally to cumulated flow shear stress. And this has been verified in all the tested soil. The key to get the linear relationship is that erosion occur when the flow is able to express a shear stress larger then pcr,b over a minimum width W0.

Validation of the equations

A set of field experiment was programmed in order to collect observations and data for gully generation. The methodology was set up in Vicarello during the 2nd year of the project. Beside the field runs, a set of experiments was conducted in the laboratory. The methodology was set up as similar as possible to the one decided for field experiments.- Then other observations were added while learning during experiments.

The data obtained showed that the relationships envisaged by the model hold well (FIG.IGES_15)

The data have been examined firstly as discussed by the KULeuven team. Width and slope are practically independent. Also W-W0 does not show any strict relationship with slope. The slope that has been used here is not the slope at which the flume was positioned but the slope inside the channel. Slope inside the channel resulted extremely variable (bot temporally and spatially) so it was used the average slope of the channel over 1.1 m where 3 control sections were positioned. Each width and depth is the average value over the three sections.

It can be stated that the degree of similarity with the field trends observed by the KULeuven is acceptable and confirm the reliability of the laboratory experiments.

W = 1.4907Q0.3243 R2 = 0.4962

Fagna2

0.25

0.2

0.15

0.1

0.05

0.1

0.15

0.2

0.25

sin a

Fagna 2

0.25

0.2

0.15

0.1

0.05

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8

int Q(m^3)

Fagna 2

0.25

0.2

0.15

0.1

0.05

0 0.0002 0.0004

0.0000 0.0000

Q (m^3/s)

0.001 0.0012 0.0000 0.0000

FIG.IGES_15:

A) Width versus the total volume of runoff (int Q; where int meant integral)

B) Width versus slope in the channel – if slope of the flume was used instead of channel slope then the data plot uniformly everywhere.

C) Width versus total discharge

W (m)

Other experiments allow the collection of a selected series of data representing the following conditions: a) erosion>>deposition or deposition in the bed negligible; and b) cross section completely included into one uniform soil layer.

Then equations (1) and (2) tell us that both D and W must vary proportionally to the time- integral of shear stress (minus critical shear stress):

W − W0 =

2kS e f





∫

 p −

 e f





pcr dt



(3)

D − D0

= k B

∫ (p − pcr

)dt

(4)

Equation (4) is valid only when removal of wall collapsed material is very quick.

Fagna2

0.14

0.12

0.1

0.08

0.06

0.04

0.02

D = 2E-06(int p) + 0.0037 R2 = 0.8823

0 10000 20000 30000 40000 50000 60000 00000

xxx x (x)

Fagna 2

0.2

0.18

0.16

0.14

0.12

0.1

0.08

0.06

0.04

0.02

10000 20000 30000 40000 50000 60000 00000

xxx x (x)

FIG.IGES_16: W-W0 versus cumulated shear (int p); D versus cumulated shear

W-Wo (m)

Under these circumstances we obtained the results shown in Fig. IGES_16, which substantially confirm the theory.






W-Wo = 3E-06(int p) + 0.0141
	R2 = 0.8709

D (m)

During the experiment flow velocity, discharge and slope values were measured and can be used to test Xxxxxxx equation during intense erosion (mobile-erodible bed). If we take hydraulic roughness as constant and we calculate velocity, we can compare it with the measured one (FIG. IGES_17).

A better agreement is obtained when the velocity is compared with total and unit discharge (FIG.IGES_18)

Consequently, from now on we will use the equation relating velocity to unit discharge fore the further developments.

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.5

1.5

Xxxxxxx' s velocity (n=1)

FIG.IGES_17: velocity versus Xxxxxxx’x.

0.7

0.6

0.5

0.4

0.3

0.2

0.1

y = 0.422x 0.3349

R 2 = 0.6466

0 0.5 1 1.5 2

discharge (L/ s)

FIG.IGES_18: velocity depends only on discharge. Slope is probably counteracted by modifications of Xxxxxxx’x n running in the opposite direction (in agreement with Xxxxxx’interpretation, see one of his paper on Earth Surface Processes).

0.7

0.6

0.5

0.4

0.3

0.2 y = 2.4782x0.3598

0.1 R 2 = 0.6051

0 0.002 0.004 0.006 0.008 0.01 0.012

unit discharge (m2/ s)

velocity (m/ s)

meas.

vel.

(m/ s)

velocity (m/ s)

Towards field conditions

In order to validate the trends obtained in the laboratory we came through a simulation based on the laboratory-validated formula and we projected the trends to the condition DW=0.1 m2 at which it is more or less considered an incision to be a gully (i.e. about one square foot). This simple relationship was used:

Q = I eff A

where Ieff is rainfall intensity minus infiltration and storage, and A is catchment area. Initial width was calculated according to this empirical equation derived from experiments:

W0 = 0.15 − 0.3sin a

(5)

From this, using our regression equations, we calculated catchment areas (A) in order to get about the threshold (0.1m2± 5%). The results are shown in the Fig. IGES_19.

GULLYER - Monte Carlo Simulations

Thesholds conditions DW=0.1 m2

10000

1000

100

measured (KUL-LEG) simulated (CNR-IGES)

0.1

slope (tan α)

FIG.IGES_19: bilogarithmic representation of catchment area vs. slope relationship.

Simulation condition:

❑ Unit disharge q (m2/s) : 0 – 0.1 (uniform distribution)

❑ Sin a : 0.03 – 0.5 (uniform distribution)

❑ ∆T (s) : neg exponential distribution – min 300 mean 900

❑ I eff (mm/h) : neg exponential distribution – min 1 mean 30

❑ Dmax = 0.5 (m)

❑ W0 = 0.15 − 0.3sin a ( m)

❑ a (sidewall erosion efficiency coefficient): : 5E-06 – 1E-05 (uniform distribution)

❑ b ( bottom erosion efficiency coefficient): : 3E-06 – 5E-06 (uniform distribution)

threshold condition are satisfied when:

− W b + W 2b2 + 0.4ab

2ab

= 9810 (0.395q 0.64 ) sina ∆T

for each threshold identified S ( critical area) is calculated with:

S = q *W * 36000000

I eff

( m2)

Threshold conditions is retained when S < 0.1

∆T

Area (m2)

WP4:

As stated in the programme of this project, a package of pedoalgorithms has to be developed. Hence we co-operated with another CNR-IGES group involved in pedofunction validation for the soil database of the Region Xxxxxx-Romagna. This allowed us to obtain some basic pedofunction such as those for Ksat and net capillary drive. We developed a Java applet (Soil Erosion Interface – SEI) which can be used directly from IGES web site.

At present only the most classical part (hydraulic pedofunctions) is fully operational.

The new part is more ambitious and aims at extending the hydrological part and adding soil surface and soil erodibility sections. Despite the new part looks much more than the already achieved part this is not true because all the buttons are already in the applet and most of the algorithms has already been calculated.

The results of the work described in this report and of the part of the job done by XXXXX, ISSDS and KULeuven will also be incorporated in SEI.

SOIL HYDROLOGY:

• Ksat (matrix), Ψ, net capillary drive.

• Ksat variation with sealing

• time to ponding

• expansion of ponded surfaces

SEI ‘s COMPONENTS

SOIL ERODIBILITY:

Interrill:

• soil detachability. Rill/gully

•

Gully

•

soil detachability

Slope/catchment size Suggestion for local modifications

SOIL SURFACE

• Random Roughness (1- decay; 2- book of pictures)

Fig.IGES_20: SEI sketch – red lines indicates parts presently under implementation.

Achievements:

1) a model for describing large part of the processes involved in interrill areas

2) a model for Ksat temporal changes

3) a model fir high-rate rill/gully erosion

4) a computer applet (Soil Equation Interface) for input value (to be completed)

Papers Submitted:

• X.Xxxxx, X. Xxxxxxxx (2001). “Equations for high rate gully erosion” (submitted, Catena, Elsevier)

• X.Xxxxxxxx, X.Xxxxx, X.Xxxxxxx. Effect of rain on the macroporosity at the soil surface. European Journal of Soil Science (accepted).

• X. Xxxxxxxx, M.P. Xxxxxxxx Xxxxxxx, M.S. Xxxxx, X. Torri. “La dinamica degli orizzonti di superficie nel bilancio idrologico del suolo” Available on the Web site of the Regione Xxxxxx Xxxxxxx. (PDF format):

xxxx://xxx.xxxxxxx.xxxxxx-xxxxxxx.xx/xxxxxxxxxxxx/xxxxx/xxx_xxxxx/xxxxxxxxx/xxx/xxxx_xxxxxxx_xxxxxxxx.xxx

Conferenceces:

Presentations for the Symposium on Gully Erosion under Xxxxxx Xxxxxx, 00-00 Xxxxx 0000, Xxxxxx, Xxxxxxx:

• Torri D. and Xxxxxxxx L. – Further equations for high-rate gully erosion

• Xxxxxxxx X., Xxxxxxxxxx X., Xxxxxxxx P., Xxxxxxxx V., Xxxxxxxxxxxx J., Xxxxxx J., Xxxxx V. and Xxxxx D. – Field experiments for gully initiation

Participation at Internetional Symposium “The significance of Soil Surface charcateristic in

soil erosion”. 20-22 sept. 2001. Strasbourg

• X. Xxxxxxxx & D. Xxxxx, M. P. Xxxxxxxx Xxxxxxx, M. S. Xxxxx Dynamics and properties of Ponding Areas. (oral Presentation).

• D. Torri.X. Xxxxxxxx, M. P. Xxxxxxxx Xxxxxxx, M. S. Xxxxx. Splash-induced Soil Surface Dynamics. (poster presentation)

Other conference presentations:

• Xxxxx D. and Xxxxxx X..P.C., Modelling within-storm soil erosion dynamics. Presented at European Climate Science Conference, Vienna, 19-23 Oct. 1998.

• Xxxxx D., Xxxxxxxx X., Xxxxxxxxx C., Xxxxxxxx Xxxxxxx M.P., Xxxxx M.S. – Soil erosion,

soil qualities and functions. Prersented at the 3rd ESSC Symposium, 27 March – 1 April 0000, Xxxxxxxx, Xxxxx.

Published:

• Torri D., Xxxxxx Xxxxx D., Xxxxxxxxxx X. and Xxxxxxxx P.(1999).Within-Storm Soil Surface Dynamics and Erosive Effects of Rainstorms, Catena, 32, 8, 131-150.

• X. Xxxxxxxx. (2001). “Capacità di invaso idrico superficiale: dinamica, misura e stima”. Riv.

di Irr. e Dren. 48(2):13-20

• X. Xxxxxxxx, X. Xxxxx, X. Xxxxxx, P. Xxxxxxxx Xxxxxxx (2001). “Effect of water quality on infiltration, runoff and interrill erosion processes during simulated rainfall. Earth Surface Processes and Landforms, 26:339-342

References

Xxxxxxxx X., 1998: Soil surface roughness dynamics and its influence on infiltration processes: experimental analysis and modelling. Ph.D thesis. dept of soil science. University of Florence. (in italian).

Xxxxxxxx D., 1996. Fluvial forms and processes. Xxxxxx, Xxxxxx Xxxxxxxx Xxxxx, Xxxxxx, 0xx impression, 218 x.

Xxxxx- Xxxxxxx X.X., Xxxxxx J.P., 1981. Extension of the Soil Conservation Service Rainfall- Runoff Methodology for Ungaged watersheed. Federal Highway Adminsitration, Environmental Division Washington D.C. Report No. 81-10.

Xxxxxxxxx A. 1998. A dynamic model of gully erosion. NATO-ASI series, Vol. 155,Springer Verlag, Berlin, 451-460.

Xxxxx R.E., Xxxxxxxx J.Y, 1978. A parameter-efficient hydrologic infiltration model. Water Resource Research. 14(3):533-538.

Whitehouse D. J., 1994. Handbook of surface metrology. Institute of Physics Publishing.

Bristol. pp. 998.

FINAL REPORT K.U. LEUVEN (PERIOD: 1ST APRIL 1998 – 30TH JUNE 2001)

1. PRESENTATION

Contractor: Responsible scientist:

Katholieke Universiteit Leuven (K.U. Leuven) Prof. Dr. Xxxx Xxxxxx

Scientific staff: Drs. Xxxxxx Xxxxxxxxxxxx

Technical staff: Xxxxxxx Xxxxx (01-01-1999 – 31-12-2000) Address: Laboratory for Experimental Geomorphology

Xxxxxxxxxxxxxx 00

X-0000 Xxxxxx, Xxxxxxx

Telephone: x00 00 00 00 00/26

Fax: x00 00 00 00 00

email: xxxx.xxxxxx@xxx.xxxxxxxx.xx.xx xxxxxx.xxxxxxxxxxxx@xxx.xxxxxxxx.xx.xx

2. MAIN WORK CARRIED OUT DURING THE REPORTING PERIOD

A) list of activities subdivided per WP and Subtask

1 Work package 1 (WP1) Within-storm changes in infiltration

1.1 Subtask 3 (ST3)

A data set on the impact of surface roughness on hydraulic roughness using flumes of variable bed roughness and slope and collected previously by K.U. Leuven and CNR has been compiled to increase the existing database.

1.2 Subtask 4 (ST4)

K.U. Leuven has been sending loess soil samples from Belgium to CNR that has been tested intensively for establishing infiltration characteristics and particularly the effect of rainwater quality on runoff and erosion response of topsoils.

1.3 Subtask 5 (ST4) Algorithm development

K.U. Leuven has provided input in the discussion on the development of a new algorithm to describe the within-storm changes in surface depression storage in relation to the decay in surface roughness.

2 Work package 2 (WP2) Development and evolution of ephemeral gullies

2.1 Subtask 1 (ST1) Field inventories

Three data sets on topographical thresholds for locations in the landscape where ephemeral gullies start and end in intensively cultivated lands in Central Belgium, have been collected by K.U. Leuven. One field campaign was conducted at the end of February and the beginning of March 1998 (table

1). During this field survey 10 ephemeral gullies, that developed during the winter season (winter gullies), were assessed. A similar field campaign was conducted one year later, i.e. March 1999. During this campaign 18 ephemeral gullies have been assessed. A third field campaign was conducted in June 1998. This campaign resulted in another 14 ephemeral gullies (summer gullies). Out of these 14 summer gullies, 5 were assessed east of Leuven and 9 west of Leuven. All 14 gullies formed during an intense rainfall event that took place on June 6th 1998, but, as shown in Table 1, total rainfall depth on this day differed significantly from area to area.

Date of survey	Number of gullies surveyed	Date of causative Rainfall event	Causative rainfall (mm) Max. Daily rainfall
Feb.- March 1998	10	2 January 1998	13.8
12 June 1998	5	6 June 1998	55.2
9-18 June 1998	9	6 June 1998	20.8 - 23.6
5-17 March 1999	18	29 November 1998	18.2 – 22

Table 1: Summary of the dates of gully survey in central Belgium and the respective causative rainfall depth. Causative rainfall depth is defined as the maximum daily rainfall depth that occurred between the last tillage pass and ephemeral gully formation.

All three data sets allow 1) to extend the comparison with formerly collected ephemeral gully threshold data sets and 2) to increase our understanding of the effect of different environmental factors on ephemeral gully erosion. An additional advantage is that for all three field campaigns, daily rainfall data from nearby stations are available for the period in which the ephemeral gullies formed.

Data collected during all three aforementioned field campaigns have also been used as input data for the Ephemeral Gully Erosion Model (EGEM) (Xxxxxx, et al., 1988). As the eroded volumes for each ephemeral gully were determined in the field, a comparison between these measured erosion volumes and the predicted erosion volumes (EGEM) could be made.

Before the start of the MWISED-project two data sets, similar to the three data sets described above, had been collected. Four ephemeral gullies were assessed at the end of winter early spring 1997 and twelve gullies were assessed during summer 1997. Table 2 summarises all parameters that have been collected for all ephemeral gullies that developed in the Belgian study area from spring 1997 till spring 1999. Each ephemeral gully was divided in as many segments as there were significant (i.e. observable by eye) changes in gully morphology along the gully profile. For each segment, length was measured by means of differential GPS, while depth and width were measured with a folding rule. Depth and width were stored as attributes of the given segment in the GPS, so that numerical and graphical information for each ephemeral gully was directly linked to each other. Given depth, width and length of each segment, ephemeral gully volume can easily be calculated by summing up the respective volumes for each segment of that ephemeral gully. Figures shown in Table 2 clearly reflect the main differences between ephemeral gullies formed at the end of winter- early spring (winter gullies) and those formed during summer (summer gullies). Summer gullies are on average very wide and shallow. They typically form after high-intensity rainfall events, removing a thin layer of freshly cultivated topsoil (seed-bed). Winter gully formation is a much slower process. First of all rainfall intensity in winter is generally much lower than in summer. Ephemeral gullies therefore, develop only after soils got sealed and crusted and infiltration capacity of the topsoil has decreased so much that even a small rainfall event causes enough runoff to create a gully. The resulting gully is rather small while its depth is generally

ENV4-CT97-687 MWISED FINAL REPORT

Parameter	Data collection method	Winter Gullies (n= 32)		Summer Gullies (n= 26)
Parameter	Data collection method	Mean value	Standard deviation	Mean value	Standard deviation
Mean gully length (m)	GPS	246	325.5	124	72.8
Mean gully width (m)	Tape measure	0.53	0.15	3.07	1.02
Mean gully depth (m)	Tape measure	0.26	0.11	0.09	0.06
Mean gully volume (m³)	Calculated	70.55	34.88	39.49	24.25

Drainage Area (m²)	GPS (differential)	61 671	90 479	14 531	10 035
Watershed length (m)	MapInfo analysis	409.4	320.7	202.7	98.7
Concentrated flow length (m)	GPS	246.4	325.5	123.9	72.8
Watershed slope (%)	Clinometer	5.2	2.1	5.2	2.2
Concentrated flow slope (%)	Clinometer	5.5	2.7	6.9	1.9
Curve number	Field observation	84.6	3.0	81.0	6.4
Soil class	Soil sample / granulometric analysis	Silt: 3%, Silt Loam: 97%		Silt Loam: 100%
Channel erodibility factor (s-1)	Auto generated	0.27	0.01	0.23	0.09
Critical shear stress (N/m²)	Auto generated	0.99	0.149	1.11	0.195
Maximum depth (m)	Field measurement	0.48	0.41	0.12	0.07
Bulk density (kg/m³)	Literature (Vandaele, 1996)	1300		1300
Particle diameter (mm)	Auto generated	0.032		0.032
Particle specific gravity (kg/m³)	Auto generated	2.62		2.62
Xxxxxxx N	Auto generated	0.03		0.03
Rain distribution type	Rainstorm distribution analysis	II		II
24 hour rainfall depth (mm)	Rain gauge and/or personal commu- nication	Max daily: 17.2 8 mm threshold: 59.4 10 mm threshold: 33.8	3.0 26.0 18.6	Max daily: 33.1	11.8
Tillage practice	Field observation	Total area tilled: 100%		Total area tilled: 100%

Table 2: Summary of the EGEM-input parameters and the way they are collected, for all gully surveys. A distinction is made between gullies formed at the end of winter-early spring (winter gullies) and those formed during summer (summer gullies).

Partner’s report: KULEUVEN 3

limited by the plough pan. (+ 30 cm). Differences in the type of “causative rainfall event” are also illustrated by the fact that a typical drainage area for winter gullies is a couple of times larger than for summer gullies (Table 2). There are of course many exceptions to this general scheme. For example, when a more erodible soil horizon (decalcified loess [C1] or calcareous loess [C2]) outcrops, winter gullies as well as summer gullies tend to be much deeper than the mean depth value.

Parallel to the field inventories for ephemeral gullies on loess-derived soils in central Belgium, ephemeral gully field inventories have also been conducted in the Alentejo (SE Portugal) and the Guadalentin (SE Spain). Collection of these data was done within the framework of another project, and has been described in Nachtergaele et al. (2001). The Mediterranean data sets were used to support and extend the results found for ephemeral gullies on loess- derived soils in central Belgium.

Results related to the data collected in both central Belgium and the Mediterranean study areas, are presented and discussed under Subtask 3 (ST3) algorithm development.

2.2 Subtask 2 (ST2) Flume experiments

2.2.1 Laboratory experiments

Sampling site A

Aba0

Aba1

Sampling site B

AbB

Abp(c)

Abp

Acp

Afp

A-horizon Colluvium

E-horizon Alluvium

Bt-horizon

Figure 1: Schematic illustration of soil types that are typically found along a catena in the Belgian loess belt (after Dudal, 1955) and of the topographical position of sampling site A and B. For cultivated areas topsoil (0.3 m) is homogenized.

Legend of soil types indicated (according to F.A.O. et al., 1998)

Aba0: Silt loam soil with argic horizon and thick A-horizon (> 40 cm) (Xxxxxx Luvisol). Aba1: Silt loam soil with argic horizon and thin A-horizon (< 40 cm) (Xxxxxx Luvisol). AbB: Silt loam soils with argic or with cambic horizon (association of truncated Luvisols and Cambisols).

Abp(c): Colluvial soil on silt loam (Eutric regosol). Suffix (c) is used when colluvial layer is thin.

Acp: Poorly drained soils on silt loam (Eutric regosol). Afp: Very poorly drained soil on silt loam (Eutric gleysol).

In order to provide a tool to accurately describe spatial and temporal variations in soil erodibility on loess-derived soils, a series of concentrated flow detachment experi-ments have been conducted. Four different soil horizons (Figure 1 and Table 2), typical for loess- derived soils in Belgium, were sampled seven times during one year.

	Sampling	Clay	Silt	Sand	BD	OM	CaCo3
Sampling Soil	depth	(%)	(%)	(%)	(kg m-3)	(%)	(%)
site horizon	(m)	0-2 µm	2-50 µm	50-2000 µm	(n=3)	(n = 4)	(n=5)
Ap	0.15	12.8	79.8	7.4	1500	2.4	-
A Bt	0.40	20.7	75.3	4.0	1530	-	-
C1	1.80	15.8	79.7	4.5	1440	-	-
C2	2.20	10.4	86.5	3.1	1390	-	16.6
B C2	0.40	8.7	86.9	4.4	1490	-	16.6

Table 2: Texture, dry bulk density (BD), organic matter content (OM) and CaCo3 content of the soil horizons under study. Location of sampling site A and B is indicated in Figure 1.

A representative range of initial soil moisture contents was thus obtained for each of the horizons under study. Undisturbed soil samples were subjected to five different combinations of slope gradient and concentrated flow discharge (Table 3).

Slope

Discharge (10-5 m3 s-1) Velocity (m s-1) Shear stress (Pa)

(m/m) Average Range Average Range Average Range

I 0.10 9.4 9.1 - 9.5 0.55 0.54 - 0.56 1.63 1.61 - 1.65

II 0.20 9.3 9.0 - 9.7 0.69 0.67 - 0.70 2.59 2.54 - 2.65

III 0.35 9.4 8.8 - 9.6 0.81 0.79 - 0.83 3.72 3.61 - 3.79

IV 0.20 19.3 19.0 - 19.6 0.93 0.92 - 0.95 3.88 3.83 - 3.91

V 0.35 19.1 17.0 - 19.5 1.09 1.04 - 1.11 5.55 5.23 - 5.62

Water temperature (°C) Xxxxxxxx number Froude number

	Average	Range	Average	Range	Average	Range
I	15.5	12.0 - 19.0	805	710 - 883	4.3	4.3 - 4.4
II	16.1	13.0 - 20.0	817	722 - 914	6.0	6.1 - 5.9
III	15.8	11.0 - 20.0	817	712 - 924	6.6	6.5 - 6.7
IV	15.3	11.0 - 20.5	1628	1451 - 1870	7.7	7.5 - 7.8
V	15.0	11.0 - 20.0	1567	1319 - 1747	8.4	8.2 - 8.6

Table 3: Flow characteristics for each of 5 combinations of slope and flow discharge that are used for testing soil erodibility for four different soil horizons.

Table 4 summarizes all experimental results. Comparing soil detachment rates (Dr) listed along horizontal lines, illustrates the relation between applied flow shear stress (τ) and Dr for a given soil horizon at a given moment. Along vertical lines, the evolution of Dr with time for each of the considered soil horizons and for a given value of τ can be found. While Table 4 gives a complete overview of the results of the flume experiments, exploring all information embedded in these results requires a fragmented and more graphical representation.

Soil Sampling

Mean GMC (kg kg-1) Dr (kg m-2s-1) at τ (Pa)...........

horizon	Date	Average	Range	1.63	2.59	3.72	3.88	5.55
Ap	7-Nov-98	17.63	16.97 - 18.25	0.00	0.00	0.00	0.00	0.00
	15-Xxx-99	18.17	16.77 - 18.91	0.00	0.00	0.00	0.00	0.01
	8-Mar-99	18.89	17.78 - 21.07	0.00	0.00	0.00	0.00	0.00
	30-Apr-99	15.38	14.79 - 16.02	0.01	0.01	0.03	0.01	0.04
	2-Jun-99	12.61	11.46 - 13.94	0.01	0.03	-	0.03	0.16
	25-Jun-99	11.45	9.89 - 13.00	0.01	0.08	0.11	-	0.15
	3-Aug-99	9.63	6.72 - 11.36	0.01	0.05	0.09	0.15	-
Bt	7-Nov-98	17.62	16.37 - 19.11	0.00	0.00	0.01	0.01	0.00
	15-Xxx-99	17.46	16.85 - 18.23	0.00	0.00	0.00	0.00	0.01
	8-Mar-99	17.81	16.78 - 18.97	0.00	0.00	0.00	0.00	0.00
	30-Apr-99	15.32	14.87 - 15.68	0.01	0.03	0.03	0.02	0.02
	2-Jun-99	13.84	12.81 - 14.43	0.03	0.12	0.10	0.10	0.11
	25-Jun-99	13.18	11.21 - 14.56	0.07	0.03	0.12	-	0.15
	3-Aug-99	13.54	11.59 - 13.28	0.03	0.09	0.01	0.04	-
C1	7-Nov-98	16.75	15.01 - 18.04	0.01	0.09	0.24	0.29	0.39
	15-Xxx-99	20.89	19.79 - 22.12	0.00	0.00	0.03	0.04	0.16
	8-Mar-99	20.39	19.11 - 21.13	0.01	0.02	0.05	0.04	0.21
	30-Apr-99	16.88	16.12 - 17.66	0.07	0.23	0.40	0.50	0.49
	2-Jun-99	14.58	12.98 - 15.43	0.09	0.28	0.35	0.61	0.56
	25-Jun-99	12.89	12.09 - 13.61	0.02	0.39	0.66	-	0.73
	3-Aug-99	13.18	12.09 - 13.74	0.04	0.24	0.65	0.67	-
C2 (2.20 m)	7-Nov-98	16.00	15.48 - 16.99	0.01	0.04	0.20	0.11	0.31
	15-Xxx-99	19.98	19.10 - 20.97	0.01	0.04	0.05	0.10	0.14
	8-Mar-99	17.67	16.92 - 18.45	0.01	0.05	0.13	0.10	0.30
	30-Apr-99	14.56	12.82 - 15.28	0.01	0.08	0.19	0.32	0.61
	2-Jun-99	13.98	12.97 - 14.94	0.01	0.08	0.19	0.20	0.48
	25-Jun-99	13.50	10.82 - 15.33	0.01	0.13	0.26	-	0.43
	3-Aug-99	12.18	5.78 - 14.74	0.01	0.13	0.34	0.41	-
C2 (0.40 m)	7-Nov-98	13.53	13.11 - 14.01	0.01	0.06	0.19	0.17	0.59
	15-Xxx-99	12.55	11.06 - 13.91	0.01	0.04	0.19	0.20	0.64
	8-Mar-99	13.16	12.85 - 13.67	0.01	0.14	0.18	0.27	0.58
	30-Apr-99	10.36	8.34 - 12.04	0.02	0.07	0.39	0.51	0.88
	2-Jun-99	6.91	4.98 - 9.47	0.02	0.16	0.30	0.45	0.63
	25-Jun-99	6.46	5.05 - 8.12	0.01	0.18	0.59	-	0.62
	3-Aug-99	4.93	3.49 - 7.65	0.02	0.06	0.26	0.54	-

Table 4: Soil detachment rate (Dr) as a function of flow shear stress (τ) and gravimetric soil moisture content (GMC) for 4 different soil horizons. C2 (2.20 m) and C2 (0.40 m) are identical horizons, sampled at different depths (see Table 2). Experimental runs were duplicated so that τ-values are mean values. If only one experimental run could be conducted, the corresponding τ-value is given in italics. The number of GMC measurements used to calculate mean GMC-values ranges between six and ten.

When subdividing the data according to soil horizon, it appears that for a given horizon, variations in detachment rate could be very well related to temporal variations in initial soil moisture content (Figure 2). When subdividing the data according to initial soil moisture content, it appears that for a given soil moisture content the ploughed topsoil horizon (Ap) and the underlying clay enriched horizon (Bt), were at least five times less erodible than the decalcified loess horizon (C1) or the calcareous loess horizon (C2) (Figure 3).

8 March 1999

0.8

0.6

-2 s-1)

Dr (kg m

0.4

0.2

0.0

0 1 2 3 4 5 6

2 June 1999

0.8

0.6

Dr (kg m-2 s-1)

0.4

0.2

0.0

0 1 2 3 4 5 6

τ (Pa)

Ap-horizon Bt -horizon C1-horizon

C2-horizon (2.20 m)

C2-horizon (0.40 m)

Linear regression on C2-horizon (0.40m)

Linear regression on Bt-horizon

Figure 2: Detachment rate (Dr) as a function of applied flow shear stress (τ) for four different soil horizons, with one horizon (C2) sampled at two different depths. Sampling depth of the respective horizons is indicated in Table 2.

(A) Soil samples used were taken on March, 8, 1999, (B) samples used were taken on June, 2, 1999.

0.25

0.20

0.15

0.10

0.05

0.00

-0.05

Ap-horizon

Bt-horizon

C -horizon

C2-horizon (2.20 m)

C2-horizon (0.40 m) Rainfall depth Sampling date

Daily rainfall depth (mm)

GMC (kg kg-1)

Oct-98

Nov-98

Dec-98

Xxx-99

Feb-99

Xxx-00

Jun-99

Jul-99

Aug-99

Sep-99

Oct-99

2.0

1.5

-1 -2

Dr (kg m s )

1.0

0.5

0.0

Figure 3: (A) Evolution of soil moisture content (GMC) of four soil horizons typical for the Belgian loess belt during one year (Nov-98 – Oct-99). Every symbol represents a soil moisture measurement (n

= 2). The line for the C2-horizon (2.20 m) is interrupted twice, since at two moments in time ground water level did not allow soil moisture measurements. For a description of the respective soil horizons: see Table 2. Note that the C2-horizon was sampled at two different depths (and consequently at two different sites). A solid line represents daily rainfall depths. The dotted vertical lines indicate 7 moments when undisturbed soil samples were taken for flume experiments. (B) Evolution of detachment rate (Dr) of four soil horizons typical for the Belgian loess belt during one year (Nov-98 – Oct-99).

Dr-values in Figure 3 are calculated for an average shear stress value of 10 Pa, using the equations:

Dr = k r

τ + b

(1)

k r = n GMC2 − m GMC + p

(2)

where kr = the erodibility parameter as defined by Xxxxxx et al (1995) and Xxxxxxx et al. (1995) (s m-1), GMC = initial gravimetric soil moisture content (kg kg-1) and n, m, p and b are regression constants (Table 5).

Soil

Regression coefficients GMC (kg kg-1)

horizon	n	m	p	b		β	ω
Ap	1.496	0.737	0.09	-0.016	0.83	0.067	0.211
Bt	-1.422	-0.025	0.036	~0	0.47	0.112	0.191
C1	4.495	3.154	0.515	-0.135	0.86	0.121	0.221
C2	-2.857	-0.056	0.18	-0.243	0.83	0.027	0.21

Table 5: Numerical input for Equations 1 and 2 for each of the four soil horizons under study. β and ω are respectively the lower and the upper limit of the initial gravimetric moisture content (GMC).

The potential of Equations 1 and 2 to predict Dr-values for soil horizons typical for loess- derived soils, is illustrated by Figure 4.

1.2

1.0

-2 -1

Dr, measured (kg m s )

0.8

0.6

0.4

0.2

0.0

0.0 0.2 0.4 0.6 0.8 1.0 1.2

D , predicted (kg m-2 s-1)

Ap-horizon (n = 62) Bt-horizon (n = 69) C1-horizon (n = 68) C2-horizon (n = 131)

y = x, MEF = 0.87

Figure 4: Predicted versus measured detachment rates (Dr) for four different soil horizons. Dr-values were predicted by Equations 1&2. For each soil horizons the parameters in Equations 1&2 (n, m, p and b; Table 5) have been obtained through non-linear regression analysis. The model efficiency statistic (MEF; Xxxx and Xxxxxxxxx, 1970) indicates how well the line of perfect agreement describes the observed variation in Dr.

Combining knowledge on spatial distribution of soil profiles and initial soil moisture content is the key to explain observed spatial and temporal variations in resistance to ephemeral gully erosion for loess-derived soils. Therefore the knowledge and relations presented above should be included in an ephemeral gully erosion routine.

From equations 1 and 2 the critical shear stress value for a given soil horizon and a given initial soil moisture content can be calculated using the following equation:

τ = −k r

(3)

where τc = critical shear stress (Pa). Equation 3 can be solved through Equation 2 to calculate kr and Table 5 that lists the regression constants for the respective soil horizons of a loess-derived soil. Critical shear stress obtained through Equation 3 represents the intrinsic critical shear stress of a given soil horizon at a given initial soil moisture content. This critical shear stress is in fact related to the initiation of particle motion under the given circumstances. A study related to the actual critical shear stress values for the initiation of ephemeral gullies is presented below (seer 2.2.1 Field experiments).

2.2.1 Field experiments

K.U. Leuven has participated intensively in the discussion on the experimental set-up for field experiments on concentrated flow erosion. With respect to the practical implementation of the experimental set-up an excursion to an experimental station in Italy (i.e. Vicarello) was made. Results obtained during the collaborative experiment are used in WP 2, ST3.

Based on field data, shear stress calculations (Figure 5) have been made for the initiation of ephemeral gully erosion in cultivated fields in SE Portugal and central Belgium. Due to logistic limitations no field experiments on ephemeral gully erosion were conducted.

Flow shear stress

τ = ρ g R Ss

ρ = water density (kg m-3)

g = accelaration due to gravity (9.81 m s-2) Ss = sinus of soil surface slope

Measured

Hydraulic radius

Flow velocity

R = (W D) / (W + 2D)

= (Q W) / (2Q + U W²)

Q = flow discharge (m3 s-1)

W = width of flow (m) D = depth of flow (m)

Calculated Measured

U = (q2/5 S3/10) n-3/5 q = unit flow discharge (m2s-1)

n =Xxxxxxx’x roughness coefficient

Calculated

Figure 5: Illustration of flow shear stress calculation. With respect to peak flow shear stress values at ephemeral gully heads in SE Portugal and central Belgium, input parameters required to solve the respective equations were derived from reference tables unless indicated otherwise. To obtain peak flow shear stress-values, calculated flow discharges should be peak flow discharges.

Calculated flow shear stress values represent the situation at the ephemeral gully head at the time of ephemeral gully initiation. These shear stress values are considered to be a good

estimator of critical shear stress (τc) for ephemeral gully initiation in the given study area. Runoff discharges at the ephemeral gully head were obtained through the hydrology component of EGEM (Xxxxxxxx, 1999), while surface slope at the gully head and width of flow, i.c. ephemeral gully channel bottom width, at the gully head have been measured in the field. Other required input parameters were derived from reference tables (Figure 5).

Results presented in Figure 6 and Table 6 show clear differences in τc for ephemeral gully initiation in the two study areas. Within the Alentejo study area ephemeral gullies were observed under three types of land use: (1) wheat fields (n = 23), (2) fields under temporal fallow (n = 12) and (3) abandoned fields (n = 5). Average τc for ephemeral gully initiation in wheat fields and temporally fallow fields were almost identical, i.e. 42 and 41 Pa respectively, while average τc for abandoned fields was somewhat higher, namely 57 Pa. However, due to the small number of observations

60 100

Cumulative percentage of ephemeral gullies (%)

Percentage of ephemeral gullies (%)

30 50

0 < x <= 5

5 < x <= 10

10 < x <= 15

15 < x <= 20

20 < x <= 30

30 < x <= 50

50 < x <= 70

70 < x <= 90

0 0

Shear stress at gully head (Pa)

Ephemeral gullies in th Alentejo, SE Portugal (n = 40) Winter gullies in central Belgium (n = 20)

Summer gullies in central Belgium (n = 13)

Cumulative percentage of ephemeral gullies initiated at a given flow shear stress in the Belgian loess belt.

Cumulative percentage of ephemeral gullies initiated at a given flow shear stress in the Alentejo, SE Portugal.

Figure 6: Distribution of peak flow shear stress at ephemeral gully heads (x.x.xxxxxxx cross- section > 930 cm2) observed in the Belgian loess belt (n = 33) and in the Alentejo, SE Portugal (n = 40). Shear stress are calculated according to procedures shown in Figure 5.

and a relatively high variation, average τc for gully initiation in abandoned fields, is not significantly different (P > 0.05) from τc in fields under wheat and/or temporal fallow. Therefore the τc-values for ephemeral gully initiation in the Alentejo are treated as one sample. For the Belgian loess belt a distinction between summer and winter gullies has been

made. Yet , with respect τc-values both gully types of the Belgian loess belt are not significantly different at the 5% level (τc winter gullies = 13 Pa and τc summer gullies = 15 Pa). Similar τc-values for winter and summer gullies were expected, since soil material and land use are very similar for both gully types, i.e. (freshly) cultivated fields on loess-derived soils. From Figure 6, however, it is clear that the distribution of τc-values over the considered shear stress classes for winter and summer gullies is not identical. While 75% of the winter gullies initiated at a τc-value between 5 and 15 Pa, summer gullies initiated at τc- values between 3.3 and 30 Pa. This may be explained as if summer gullies can initiate at low shear stress values (3-4 Pa), but due to high(er) rainfall intensities in summer, large discharges and consequently large shear stress values do occur as well.

Generally it is clear that average τc for ephemeral gully initiation in the Belgian loess belt (14 Pa) is significantly different (P < 0.01) from average τc for the Alentejo (44 Pa). For both study areas mainly cultivated fields were considered, but the most important difference lies in the rock fragment content of the topsoil. Whereas no rock fragments occur in Belgian loess- derived (top)soils, rock fragment content of topsoils in the Alentejo amounts to 30% by mass on average. Xxxxxx et al. (1999) experimentally developed a negative exponential relation between rock fragment cover and sediment concentrations in concentrated flow. They also found that the erosion reducing effect of rock fragments on concentrated flow erosion rates was especially significant in the case of initially wet topsoils, a precondition that can easily be fulfilled for the Portuguese study area, since ephemeral gullies considered there developed during Mediterranean winter.

In conclusion, results of this shear stress threshold analysis reveal clear differences between required concentrated flow shear stress values for the initiation of ephemeral gullies in the Alentejo and in the Belgian loess belt. This information can for example be used to set up experimental programmes for ephemeral gully erosion simulations under the respective circumstances. But τc-values as presented here should also be used by ephemeral gully erosion models to define critical thresholds for ephemeral gully initiation and/or to spatially limit zones of potential ephemeral gully erosion.

Table 6A

Alentejo, SE Portugal

Table 6B

Belgian loess belt, winter gullies

Runoff Width Hydraulic		Sine of	Shear		Runoff Width Hydraulic Sine of Shear
Discharge	radius	slope	stress	Discharge	radius	slope	stress

3 -1

(m s ) (m) (m) (Pa)

0.0057 0.80 0.01 0.16 16.8

0.0085 1.20 0.01 0.21 20.6

0.0142	0.65	0.02	0.10	22.3
0.0057	0.50	0.01	0.18	23.4
0.0085	1.10	0.01	0.30	27.1
0.0227	0.60	0.03	0.09	27.7
0.0255	0.75	0.03	0.10	29.8
0.0680	0.20	0.06	0.05	29.9

0.0142

0.80

0.02

0.18

30.9

0.0028

0.40

0.01

0.08

9.8

0.65	0.02	0.17	33.1
0.55	0.02	0.19	33.6

0.0142

0.0000

0.0000 0.65 0.02 0.15 33.6

0.40	0.02	0.19	33.9
0.50	0.02	0.22	34.2

0.0085

0.0000

0.0000 0.60 0.03 0.13 36.1

0.55	0.06	0.06	36.7
0.45	0.03	0.12	38.3

0.0680

0.0000

0.0000 0.60 0.03 0.13 38.5

	0.0142	0.50	0.02	0.19	39.7
	0.0170	0.60	0.02	0.19	40.3
	0.0340	0.45	0.04	0.10	41.0
	0.0453	0.65	0.04	0.10	41.8
	0.0198	0.70	0.02	0.20	42.2
0.0198 0.60			0.02	0.18	42.3
0.0227 0.60			0.03	0.16	42.8
0.0113 0.25			0.03	0.18	45.2

3 -1

(m s ) (m)	(m)		(Pa)
0.0028 0.40	0.02	0.02	3.6

0.0028 0.50 0.01 0.04 5.3

0.0028		0.40	0.02	0.04	6.0
	0.0028	0.38	0.01	0.05	7.2
	0.0028	0.58	0.01	0.07	7.3
	0.0113	0.60	0.03	0.03	8.6
	0.0057	0.50	0.02	0.05	9.2
	0.0028	0.35	0.01	0.07	9.5

0.40	0.01	0.08	9.8
0.50	0.01	0.10	10.2

0.0028

0.0000

0.0000 0.45 0.01 0.10 10.8

0.52	0.01	0.11	11.0
0.34	0.01	0.09	12.0

0.0028

0.0000

0.0000 0.35 0.01 0.10 12.3

0.40	0.03	0.05	12.7
0.60	0.03	0.07	17.9

0.0085

0.0000

0.0000 0.37 0.03 0.09 26.2

0.0113		0.40	0.02	0.14	31.3
	0.0198	0.50	0.03	0.11	32.2
mean	0.0061	0.45	0.02	0.07	12.6
St. dev.	0.0052	0.08	0.01	0.03	8.1

Table 6C

Belgian loess belt, summer gullies

Runoff Hydraulic Sine of Shear

0.0170 0.45 0.03 0.19 46.0

0.0680 0.55 0.06 0.08 47.9

Discharge

3 -1

Width

radius slope stress

	0.0227	0.40	0.03	0.17	51.6		(m s )	(m)	(m)		(Pa)
	0.1133	0.60	0.07	0.07	54.0		0.0028	4.00	0.003	0.11	3.3
	0.0283	0.25	0.04	0.14	55.0		0.0028	2.50	0.004	0.08	3.6
	0.0481	0.65	0.04	0.14	55.1		0.0028	1.20	0.01	0.11	6.7
	0.0453	0.60	0.04	0.15	57.9		0.0283	3.00	0.02	0.05	8.8
	0.0708	0.45	0.06	0.11	62.2		0.0085	1.50	0.01	0.11	11.2
	0.0255	0.50	0.03	0.23	64.3		0.0085	1.80	0.01	0.15	12.5
	0.0651	0.35	0.06	0.13	70.7		0.0198	1.00	0.03	0.06	15.0
	0.0312	0.20	0.04	0.18	71.3		0.0227	0.70	0.04	0.04	15.6
	0.0708	0.50	0.05	0.14	72.9		0.0453	1.40	0.04	0.05	19.0
	0.0623	0.45	0.05	0.15	74.2		0.0538	1.80	0.03	0.07	21.8
	0.0821	0.60	0.05	0.14	74.4		0.0283	1.40	0.02	0.10	22.2
mean	0.0326	0.56	0.03	0.15	43.5		0.0510	1.70	0.03	0.09	26.0
St. dev.	0.0258	0.20	0.02	0.05	15.7		0.0368	1.20	0.03	0.11	29.9
						mean	0.0240	1.78	0.02	0.09	15.1
						St. dev.	0.0185	0.90	0.01	0.03	8.4

Table 6: Measured and calculated flow parameters at ephemeral gully heads in Alentejo, SE Portugal (Table 6A, n = 40), in the Belgian loess belt for winter gullies (Table 6B, n = 20) and for summer gullies (Table 6C, n = 13). Xxxxx stresses listed served as input data for Figure 6.

2.3 Subtask 3 (ST3) algorithm development

One goal of MWISED is to develop a sub-model to predict where and when ephemeral gullies will occur. In order not to have to start from scratch the Ephemeral Gully Erosion Model (EGEM), which is a physically-based model that was specifically developed to predict soil loss by ephemeral gully erosion in North America, was first tested for European ephemeral gullies (central Belgium, SE Spain and SE Portugal). The model has two major

components, of which the hydrology component is a physical process model, based on the runoff curve number. The erosion component uses the hydrology outputs to solve a combination of empirical relationships and physical process equations in order to compute the final width and depth of the ephemeral gully (Xxxxxxxx, 1999). Results of testing EGEM clearly showed that ephemeral gully cross-sections observed in the Mediterranean study areas are overpredicted, while for the Belgian loess belt mean ephemeral gully cross- sections of both winter and summer gullies are underpredicted. (Figure 7). It was concluded that the physically-based erosion technology as included in EGEM, does not yield satisfying results with respect to predicting ephemeral gully cross-sections. Moreover, no routine to predict the length or the location of an ephemeral gully is included in EGEM which is also considered to be a major limitation.

1.0

Mean measured ephemeral gully cross-section (m²)

0.8

0.6

0.4

0.2

0.0

0.2 0.4 0.6 0.8 1.0

Mean predicted ephemeral gully cross-section (m²)

Alentejo (n = 40) guadalentin (n = 46)

Belgian loess belt, winter gullies (n = 32)

Belgian loess belt, summer gullies (n = 26) Line of perfect agreement

Figure 7: Predicted versus measured mean ephemeral gully cross- sections for two Mediterranean study areas (Alentejo, SE Portugal and Guadalentin, SE Spain) and for summer and winter gullies in the Belgian loess belt. Ephemeral gullies were predicted by XXXX.

With respect to the development of an alternative ephemeral gully erosion routine within the MWISED project (Figure 8) K.U. Leuven provided crucial input (data, equations and algorithms). Three main issues have been addressed (black ellipses in Figure 8): 1) delineate zones prone to ephemeral gully erosion, 2) link the hydrology component to the erosion component and 3) develop a dynamic approach of the soil erodibility concept.

HYDROLOGY

EROSION

TOPOGRAHY

INPUT

Runoff con- trib. area

Surface slope

PROCESSES AND FACTORS CALCULATED WITHIN A GRIDCELL

Zones prone to gully erosion

KINEMATIC WAVE FOR TRANSPORT BETWEEN CELLS

Sediment discharge in gullies

OUTPUT

Erosion/ Sedimentation

Flow Erosion

Transport Capacity

Gully width

Conc. over- land flow

topographic threshold

Splash erosion

Infiltration

DEM

Rainfall

Local Drain Direction

New DEM

Water discharge in gullies

Figure 8: Scheme of the ephemeral gully erosion routine to be implemented in a MWISED (after van de Vlag et al., 2000). Black ellipses indicate newly developped model elements.

Delineate zones prone to ephemeral gully erosion

The procedure to determine zones prone to ephemeral gully erosion can be split up in a procedure to determine where ephemeral gullies will start and where they will end and a procedure to route the water between these two points. Initiation of ephemeral gullies has been described in literature as a threshold phenomenon. Xxxxx (1966) and Xxxxxx (1973) were amongst the first to publish data on how the relation between runoff contributing area

(A) and slope of soil surface (S) at gully heads may be used to establish a topographical threshold relation for gully initiation. Vandaele et al. (1996) summarized the available information on the initiation and location of (ephemeral) gullies, also reporting SA-relations for the Belgian loess belt, and Vandekerckhove et al. (1998) focused on the potential of the topographical threshold concept for predicting ephemeral gully initiation points in Mediterranean areas.

With respect to points in the landscape where ephemeral gullies end, much less research has been conducted. An attempt to establish a topographical threshold relation for sediment deposition points in the Alentejo and Guadalentin study area is presented in Figure 9. Although a profound study of such threshold relations needs more data from more different environments, it is possible to draw some significant regressions for both study areas. Both relations are significant at the 5% level. From Figure 9 it can be seen that ephemeral gullies in the Guadalentin form sedimentation fans on much steeper slopes than is the case in the Alentejo. This can partly be attributed to the coarser sediment load (rock fragments) transported in the ephemeral gullies of the Guadalentin.

Sediment deposition in the Belgian loess belt appeared to be better predicted by a slope threshold approach instead of a S-A-relation. All sediment deposition points

Log y = -1.00 - 0.20 log x y = 0.10 x-0.20, R2 = 0.42

Log y = -1.48 - 0.54 log x y = 0.03 x-0.54, R2 = 0.63

Surface slope (m m-1)

0.1

0.01

0.001 0.01 0.1 1 10

Down area (ha)

Ephemeral gully sediment deposition, Guadalentin (n = 37) Ephemeral gully sediment deposition, Alentejo (n = 20)

Figure 9: Topographical thresholds for ephemeral gully sediment deposition points in the Guadalentin (SE Spain) and the Alentejo (SE Portugal). Downarea is the runoff contributing area at the gully fan, i.e. sedimentation point.

assessed in the Belgian loess belt were essentially controlled by a change in topography (i.e. a decrease of local slope) and not by a change in land use. For about 90% of the ephemeral gullies in the Belgian loess belt channels end by sediment deposition at slopes of 4% or less (Figure 10). When the lowermost gully points have to be determined from a DEM for example, a slope threshold of 3% seems to be a good estimator.

Whatever method is used, when the initiation point and the sediment deposition point of an ephemeral gully are known, ephemeral gully length can be derived by routing the water from this initiation point (ephemeral gully head) towards the ephemeral gully end. Xxxxxx et al. (1999) showed that zones prone to ephemeral gully erosion can be successfully predicted from a topographical threshold concept.

Once the length of the concentrated flow zone or the expected ephemeral gully is known, the ephemeral gully volume can be directly derived using an equation of the form:

Veg = a Lb

(4)

where Veg = ephemeral gully volume (m3) and Leg = ephemeral gully length (m). For the ephemeral gullies in the Mediterranean study areas and for the summer gullies in the Belgian loess belt, a = 0.048 and b = 1.29 (n = 112, R2 = 0.91). For winter gullies in the Belgian loess belt, a = 0.060 and b = 1.15 (n = 31, R2 = 0.82) (Figure 11).

12 1.00

Cumulative Frequency (%)

0.75

Frequency

6 0.50

0.25

x <= 0.01

0.01< x <= 0.02

0.02< x <= 0.03

0.03< x <= 0.04

0.04< x <= 0.05

0.05< x <= 0.06

0.06< x

0 0.00

Sediment deposition slope (m m-1)

Figure 10: Distribution of slopes where ephemeral gullies end by topographically controlled sediment deposition as measured in the Belgian loess belt for winter gullies and summer gullies. (n = 39)

1000

Ephemeral gully volume (m³)

100

0.1

10 100

1000

Ephemeral gully length (m)

Eph. gullies in Mediterranean study areas (n = 86)

Ephemeral gullies in Xxxxxx loess belt, summer gullies (n = 26) Ephemeral gullies in Xxxxxx loess belt, winter gullies (n = 31) Power regression on Mediterranean and summer gullies (n = 112) log y = -1.32 + 1.29 log x, or y = 0.048 x 1.29, R 2 = 0.91

Power regression on winter gullies (n = 31)

log y = -1.22 + 1.15 log x, or y = 0.060 x 1.15, R 2 = 0.82

Figure 11: Potential performance of an empirical regression model relating ephemeral gully length and ephemeral gully volume on a double logarithmic scale.

When a more physically-based approach is envisaged, ephemeral gully volumes need to be calculated through process equations describing the detachment process. In this case, a hydrology component can be used to calculate (peak) flow discharge for each cell within the zones classified as ‘prone to ephemeral gully erosion’. Yet, to transform this peak discharge into erosive power of the concentrated flow (e.g. shear stress), flow width has to be known (Figure 8, black ellipse number 2).

Link the hydrology component to the erosion component

Empirical prediction equations of the form W = a Qb have been reported for rills and rivers, but not for ephemeral gullies. Therefore, six experimental data sets are used to establish a channel width (W, m) – flow discharge (Q, m3 s-1) relation for ephemeral gullies formed on cropland. The resulting regression equation (W = 2.51 Q0.412; R2 = 0.72; n = 67) predicts observed channel width reasonably well (Figure 12).

Line of perfect agreement

0.7

Observed channel width (m)

values observed on cropland or simulated cultivated topsoils

0.6

0.5

0.4

0.3

0.2

0.1

0.0

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Predicted channel width (m)

values predicted by W = 2.51 Q0.412 (Equation 6.11)

Figure 12: Predicted channel width versus observed channel width. The channel width prediction equation was established through a non-linear regression on channel width – flow discharge data for rills and gullies that formed on cropland or in simulated cultivated topsoils.

ENV4-CT97-687

Table 7: Characteristics of the experimental data sets reporting on channel width and flow discharge for rills and gullies developing on cropland or simulated cultivated topsoils. Runoff discharges indicated by

* are calculated, n.a. = not available

Partner’s report:

Data set name

Number of observations

Data collection method

Topsoil

condition Slopes

Flow Discharge

Channel width (m)

Data set source

MWISED

FINAL REPORT

KULEUVEN

				(%)	(10-3m3 s-1)
Xxxxx Xxxx	4	Laboratory flume experiments (USA)	Seedbed conditions, stony topsoils	7.0 – 14.0	0.10 – 0.19	0.08 – 0.18	Xxxxx-Xxxx, 0000
Xxxxxx	10	Laboratory flume experiments (France)	Seedbed conditions, silt loams	n.a.	0.19 – 0.76	0.10 – 0.27	Xxxxxx et al., 1990
Xxxxxx	7	Laboratory flume	Seedbed conditions	8.3 – 9.9	0.30 –2.18	0.08 – 0.24	Xxxx and Xxxxxx,
		experiments (USA)					1980
Xxxxxx	n.a.	Field experiments (USA)	Seedbed conditions, different soils	5.5 – 9.8	0.02 – 1.83	0.02 – 0.27	Xxxxxx et al., 1990
Vicarello	7	Field experiments (Italy)	Cropland, clay rich soils	n.a.	1.34 – 9.30	0.13 – 0.41	Xxxxxxxx et al., 2000
Xxxxxxxxx (flume)	20	Open air flume ex- periments (Australia)	Natural soils mainly composed of silt	6.3 – 59.4	1.40 – 11.00	0.12 – 0.23	Xxxxxxxxx, 0000
Xxxxxxx	19	Open air flume experiments (USA)	Seedbed conditions, clay loam	0.6-1.0	4.40 – 28.70	0.32 – 0.59	Xxxxxxx et al., 0000
Xxxxxxxxxxxx (Xxxxxxx)	42	Field measurements (winter gullies;	Cropland, loess- derived soils	0.5 – 14.0	2.80 – 53.80*	0.25 – 1.10	Xxxxxxxxxxxx xx xx. (xx xxxxx)
		Xxxxxxx)
Xxxxxxxxxxxx	00	Field measurements	Cropland, stony top	5.0-31.0	14.16 – 540.90*	0.18 – 3.70	Nachtergaele et
(Portugal)		(Portugal)	soils				al., 2001

Due to logistic limitations related to the respective experimental set ups, only relatively small runoff discharges (i.e. Q < 0.02 m3 s-1) were covered. Using field data, where measured ephemeral gully channel width was attributed to a calculated peak runoff discharge on sealed cropland, the application field of the regression equation was extended towards larger discharges (i.e. 5 10-4 m3 s-1< Q < 0.1 m3 s-1) (Figure 13).

Line of perfect agreement

Channel width (m)

values observed on cropland or simulated cultivated topsoils

0.1

0.1 1 10

Channel width (m)

values predicted by W = 2.51 Q0.41 (Equation 6.11)

Experimental cropland data, Q measured (n = 67) Field cropland data, Q predicted (n = 112)

Figure 13: Predicted channel width versus observed channel width. ‘Experimental cropland data’ refers to the data presented in Figure 10. ‘Field cropland data’ refers to ephemeral gully data obtained by Nachtergaele et al. (2001 and in press). More information on the respective data sets can be found in Table 7.

Comparing W-Q relations for concentrated flow channels revealed that the discharge exponent (b) varies from 0.3 for rills over 0.4 for gullies to 0.5 for rivers. This shift in b may be due to 1) differences in flow shear stress distribution over the wetted perimeter between rills, gullies and rivers, 2) a decrease in probability of a channel formed in soil material with uniform erosion resistance from rills over gullies to rivers and 3) a decrease in average surface slope from rills over gullies to rivers.

The proposed W-Q equation for ephemeral gullies is valid for (sealed) cropland with no significant change in erosion resistance with depth. In the case of a typical summer situation where the soil moisture profile of an agricultural field makes the top 0.02 m five times more erodible than the underlying soil material (Figure 14), observed W values for summer gullies are larger than those predicted by the established channel width equation for concentrated flow on cropland (Figure 15).

Soil moisture content (kg kg-1)

0.00 0.04 0.08 0.12 0.16 0.20




	Observation site 1 Observation site 2 Observation site 3 Average trend A

0.00

-0.05

Depth (m)

-0.10

-0.15

-0.20

-0.25

0.00

Relative erodibility, K

1 2 3 4 5 6

-0.05

Depth (m)

-0.10

-0.15

-0.20

-0.25

Figure 14: A) Gravimetric soil moisture content profile for three observation sites in the Belgian loess belt. Time (June 2000) and location of soil moisture sampling represent the conditions of potential summer gully initiation. At each observation site two sampling points were selected and soil moisture content of each sample was analysed in duplicate. The average trend therefore, represents the average for 12 gravimetric soil moisture values. B) Relative erodibility during concentrated flow erosion (based on experimental results by Xxxxxx et al., 1990)

Line of perfect agreement

Channel width (m)

values observed on cropland

0.1

1 10

Channel width (m)

values predicted by W = 2.51 Q 0.41 (Equation 6.11)

Summer gullies - Belgian loess belt (n = 39)

Figure 15: Predicted channel width versus observed channel width for ephemeral gullies formed in central Belgium during summer. The prediction equation was established for experimental rills and gullies that formed on cropland soils with uniform erosion resitance.

Besides the development of empirical channel width (W) – flow discharge (Q) relations discussions between K.U. Leuven and CNR also resulted in a more physically based approach. Xxxx Xxxxx (CNR) developed procedures to predict W from Q, through a process-oriented approach. To investigate how this W-Q relation changes over time, an experimental procedure was elaborated and a first “try-out” experiment was conducted at Vicarello (Italy) from October 1st till October 4th, 1999. This experiment was headed by CNR and ISSDS, and attended by IIA, CSIC and K.U. Leuven. The collaboration resulted in a procedure to conduct ephemeral gully field experiments under standardised conditions.

A dynamic approach of the soil erodibility concept

Field observation showed that both spatial and temporal variations in resistance to ephemeral gully erosion for loess-derived soils exist. At a given moment in time, detachment rates for different soil horizons vary significantly (spatial variations). Soil profiles in the Belgian loess belt are truncated due to water and tillage erosion, resulting in a spatial variation in outcropping soil horizons (Figure 1) and consequent spatial variation in soil erodibility. When modelling soil erosion for loess-derived soils, the use of one single erodibility value will therefore lead to serious errors on predicted erosion volumes and patterns. Besides spatial variations, also temporal variations in soil detachment rates have been observed. For a given soil horizon, these variations could be very well related to

temporal changes in initial soil moisture content (Figure 4). Modelling concentrated flow erosion requires that both temporal and spatial variations in detachment rates should be incorporated. The level of incorporation depends on the type of erosion model that is applied. When an empirical erosion model is used, a (relative) distinction in detachability between two time periods (winter and summer) and two groups of soil horizons (Ap-Bt and C1-C2) may be sufficient (Figure 3). On the other hand, process-based physical erosion models both enable and require a more detailed approach. To fully incorporate the dynamic approach of the soil erodibility concept, such a process-based physical erosion model should be multi-layered in order to describe soil profiles within the study area. It is important to stress that the progressive availability of digital information (i.c. a digital soil map) will facilitate the development and use of model structures requiring spatially distributed input data (i.c. soil profile distribution) in the near future. From the distribution of soil profiles and the related distribution of gravimetric soil moisture content, spatial and temporal variations in detachability throughout the study area can be calculated using the results presented in Equations 1&2 and in Table 4.

A more detailed description of the development of the dynamic approach of the soil erodibility concept can be found under 2.2 Subtask 2 (ST2) Flume experiments, 2.2.1 Laboratory experiments.

2.4 Subtask 4 (ST4) GIS development

Within subtask 4 a collaboration between UU and K.U. Leuven was established Figure 8 shows the flowchart of a gully model as it was elaborated by UU and K.U. Leuven. The hydrology component and the erosion component of this model use procedures that were already incorporated in EUROSEM. The topography component results from several discussions between UU and K.U. Leuven. The topography component will delineate zones in the landscape which are potentially prone to ephemeral gully erosion. Critical threshold conditions for ephemeral gully erosion have been discussed. It was agreed with K.U. Leuven and UU to use topographical threshold procedures as they have been reported in literature by members of K.U. Leuven (Vandaele et al. 1996; Xxxxxx et al., 1999). Once these zones have been selected the kinematic wave will only be applied on cells that fall within these zones.

Besides the W-Q relationships (see 2.3 Subtask 3 (ST3) algorithm development), also some input data to test the ephemeral gully model for a small catchment in the Belgian loess belt was exchanged between K.U. Leuven and UU. K.U. Leuven delivered:

1) 2 DEM’s (1x1 m and 5x5m ) of the study area (+ 5 ha);

2) rainfall data (tipping bucket rain gauge; accuracy 0.2mm) of the rain event that created 4 ephemeral gullies in the study area;

3) saturated hydraulic conductivity (Ksat), D50 and cohesion values for the Belgian loess soils;

4) location and length, depth, width and eroded volume of 4 ephemeral gullies that were created within the study area.

B) Achievements

1) A field data set on ephemeral gully erosion in central Belgium was collected. In total 58 ephemeral gullies were assessed, 32 ephemeral gullies developed during winter or early spring (winter gullies) and 26 ephemeral gullies developed during summer (summer gullies).

2) The Ephemeral Gully Erosion Model (EGEM) was tested for all 58 ephemeral gullies that were measured in the Belgian loess belt. Results showed that EGEM is not capable of predicting ephemeral gully cross-section in this study area. An identical conclusion was drawn after testing EGEM for data set containing 82 ephemeral gullies that developed in two Mediterranean study areas (Alentejo, SE Portugal and Guadalentin, SE Spain). This data set was collected within another project. Because of EGEM’s inability to predict ephemeral gully cross-sections and because of the fact that EGEM does not allow to predict the location and/or length of an ephemeral gully, an alternative ephemeral gully erosion routine had to be developed.

3) Topographical threshold conditions to delineate zones in the landscape that are prone to ephemeral gully erosion have been determined in consultation with UU. Topographical threshold relations for ephemeral gully initiation were derived from literature. Vandaele et al. (1996) reported relations for the Belgian loess belt, while Xxxxxxxxxxxxxx et al. (1998) focused on predicting ephemeral gully initiation points in Mediterranean areas. With respect to points in the landscape where ephemeral gullies end a topographical threshold relation is established for the Mediterranean areas (Figure 9), while for the Belgian loess belt a critical slope threshold of 4% was proposed (Figure 10). When initiation and sediment deposition point are known, ephemeral gully length can be derived by routing the water between these two points following an algorithm developed by Xxxxxx et al. (1999).

4) Within an empirical modelling approach procedures described under achievement 3 can be used to directly calculate ephemeral gully volumes from the predicted ephemeral gully

length: Veg = a Lb

(Equation 4) where Veg = ephemeral gully volume (m3) and Leg =

ephemeral gully length (m). For the ephemeral gullies in the Mediterranean study areas and for the summer gullies in the Belgian loess belt, a = 0.048 and b = 1.29 (n = 112, R2 = 0.91). For winter gullies in the Belgian loess belt, a = 0.060 and b = 1.15 (n = 31, R2 = 0.82) (Figure 11).

5) Within a physically-based modelling approach procedures described under achievement 3 will be used to delineate zones prone to ephemeral gully erosion. For each point within these ephemeral gully prone areas, the erosive power can be calculated using the established width-discharge relationship that allows to link the hydrology component (discharge) to the erosion component (erosive power). The proposed relation W = 2.51 Q0.412 is valid for cropland with no significant change in erosion resistance with depth.

6) In order to accurately transform erosive power to erosion rates, flume experiments have been conducted that test differences in erodibility between different soil horizons typical for

the Belgian loess belt. These experiments show that variations in detachment rate (Dr) for a loess-derived soil can be very well predicted as a function of soil horizon type and initial soil moisture content (GMC) (Figure 4). The established equations are:

Dr = k r τ + b

k r = n GMC2 − m GMC + p

7) An analysis of critical shear stress values (τc) for ephemeral gully intiation in SE Portugal and the Belgian loess belt has been conducted (Table 6). Based on field data, τc was calculated according to procedures presented in Figure 5. Generally, τc for ephemeral gully initiation in the Belgian loess belt (14 Pa) was significantly different (P < 0.01) from average τc for the Alentejo (44 Pa). Since for both study areas mainly cultivated fields had been considered, the difference in τc was attributed to the rock fragment content of the topsoil.

3. ACTIVITY INTERNAL TO THE PROJECT.

3.1 Meetings:

1) In the first year two MWISED meetings were organized in Leuven: the start–up meeting, May 29-30, 1998 and a Gully Erosion Workshop, November 14, 1998.

2) In the second year K.U. Leuven, participated in 4 MWISED-meetings

a) Murcia April 24-28, 1999

b) Firenze/Vicarello October 1-4, 1999

c) Utrecht January 14, 2000

d) Firenze June 8-10, 2000

3) During the second year also three informal meetings related the gully modelling procedures were held in Leuven:

a) July 0000, Xxxx Xxxxx (XXX) visited K.U. Leuven to discuss possibilities of predicting cross-sectional areas and/or width and depth of flow at a given point in the landscape, from drainage area (as a substitute for discharge) and local slope. Finally procedures to predict changes of width over time as a function of discharge were elaborated by Xxxx Xxxxx.

b) December 1999, Xxxxxx xxx xx Xxxx (UU) visited K.U. Leuven to discuss the proposed outline of the ephemeral gully model component for MWISED

c) January 0000, Xxxxxx Xxxxxx and Xxxxxx xxx xx Xxxx (UU) visited K.U. Leuven to discuss preliminary results of the ephemeral gully model component for MWISED, and further improvements.

3.2 Data exchange

1) Rainfall data were exchanged with Xxxxx Xxxxxxx. Two data sets, each covering 10 years of rainfall data (Uccle, Brussels) with a resolution of 10 minutes (1934-1943 and 1985-

1994), were obtained from the Belgian Royal Meteorological Institute. The data could be obtained at no costs, under the restriction that these data will never be used for commercial purposes of any kind.

2) Input data for the ephemeral gully model as well as data to test the ephemeral gully model were delivered to UU.

4. ACTIVITY EXTERNAL TO THE PROJECT.

4.1 papers

Nachtergaele, J., Xxxxxx, J., 1999. Assessment of soil losses by ephemeral gully erosion using high-altitude (stereo) aerial photographs. Earth Surface Processes and Landforms, 24: 693-706.

Xxxxxx, X., de Xxxx, X., Xxxxxx, X., Xxxxxxxxxxxx, J., Xxxxxx, G., 1999. Concentrated flow erosion rates as affected by rock fragment cover and initial soil moisture content.

Catena, 36: 315-329.

Xxxxxxxx, X., Xxxxx, X., Xxxxxx, X., Xxxxxxx, X.X., 0000. Effects of water quality on infiltration, runoff and interrill erosion processes during simulated rainfall. Earth Surface Processes and Landforms , 26: 329-342.

Nachtergaele J, Xxxxxx J, Vandekerckhove L, Oostwoud Wijdenes D, Xxxx M. 2001.

Testing the Ephemeral Gully Erosion Model (EGEM) for two Mediterranean environments. Earth Surface Processes and Landforms , 26 (1): 17-30.

Accepted for publication in Geomorphology.

Nachtergaele, J., Xxxxxx, J., submitted. Spatial and temporal variations in resistance of loess-derived soils to ephemeral gully erosion. European Journal of Soil Science.

Nachtergaele, J., Xxxxxx, J., Xxxxxxxx Xxxxxxxx, D., Vandekerckhove, L., submitted.

Medium-term evolution of a gully developed in a loess-derived soil.

Geomorphology.

Nachtergaele, X., Xxxxxx, J., Xxxxxxxxx, X., Xxxxx, D., submitted. Flow width – discharge relations for rills and (ephemeral) gullies. Hydrological Processes.

Xxxxxxxxx, X., Xxxxxx, X., Xxxxxxxxxxxx, J., 1999. Gevoeligheid van drie lemige bodemhorizonten voor erosie door geconcentreerde afvoer. De Aardrijkskunde , 3: 11-18

Nachtergaele J., Xxxxxx X., 2000. EGEM, a potential prediction tool for soil losses by ephemeral gully erosion in the Belgian loess belt?. Xxxxxxxx, D., Schiettecatte, X. (Eds.), Proc. Erosion contact group-meeting, March 11th, 1999, International

Center for Eremology, Ghent Univeristy, Belgium. I.C.E. Special report No. 2/2000: 55-60.

Nachtergaele, J., Xxxxxx, J., Vandekerckhove, X., Oostwoud Wijdenes, D., Xxxx, M., in press. Testing the Ephemeral Gully Erosion Model (EGEM) in Mediterranean environments. Proc. 10th International Soil Conservation Organization (ISCO) Conference: Sustaining the Global Farm. Local Action for Land Stewardship. Purdue University, West Lafayette, Indiana, USA, 23-28 May, 1999.

4.2 paper presentations at conferences

Nachtergaele, J., Xxxxxx, J., 1998. Ephemeral Gully Erosion Assessment for the last 50 Years via High Altitude Stereo Aerial Photographs. Case Study: The Belgian Loess Belt. ESSC-workshop: Long-term Effects of Land Use on Soil Erosion In a historical perspective, Müncheberg, Germany, Sept. 11-13, 1998.

Nachtergaele, J., Xxxxxx, X., Xxxxxxx, A., Xxxxxx, I., Xxxxxxxxxx, X., Xxxxxxxxxxxxxx, L., Xxxxxx, X., 1998. Prediction of soil losses by ephemeral gully erosion using EGEM (ephemeral gully erosion model). Gemeenschappelijke studiedag van de Belgische verenigingen voor bodemkunde en landelijk genie “Studie van bodem en duurzame ontwikkeling”. Centre de Recherches Agronomiques, Gembloux, Belgium, November 25, 1998.

Nachtergaele, J., Xxxxxx, J., 1999. Ravijnerosie en ravijnerosie-onderzoek in de Belgische Xxxxxxxxxx. Bijeenkomst contactgroep erosie, March 11, 1999. Universiteit Gent, centrum voor eremologie, Gent, België.

Nachtergaele, J., Xxxxxx, J., Xxxxxxx, A., Xxxxxx, I., Xxxxxxxxxx, X. and Xxxxxx, G., 1999.

Ephemeral gully erosion in the Belgian Loess Belt. 2nd Int. Symposium on Tillage Erosion and Tillage Translocation. K.U. Xxxxxx, Xxxxxxx, 00-00 April 1999.

Xxxxxx, X., Xxxxxxxxxxxx, X., Xxxxxxxxxxxxxx, X., Xxxxxxxx-Xxxxxxxx, X., 0000.

Datasets needed for predicting ephemeral gully erosion under global change. BGRG Rainfall Simulation Working Group Concluding Meeting (Joint Meeting with COST 623 Soil Erosion and Global Change). 18-21 April 1999, Leicester, U.K.

Nachtergaele, J., Xxxxxx, X., Vandekerckhove, X., Oostwoud Wijdenes, D., Xxxx, M., 1999. Testing and evaluating the ephemeral gully erosion model (EGEM) in Mediterranean environments. 10th International Soil Conservation Organization (ISCO) Conference: Sustaining the Global Farm. Local Action for Land Stewardship. Purdue University, West Lafayette, Indiana, USA, May 23-28, 1999.

Xxxxxx, X., Xxxxxxxxxxxx, X., Vandekerckhove, X., Oostwoud Wijdenes, D., Xxxx, M., 1999. Prediction of ephemeral gully erosion in Mediterranean environments. I.A.G. Regional Conference on Geomorphology. University of Rio de Janeiro, Brazil, July 17-22, 1999.

Nachtergaele, J., Xxxxxx, X., Xxxxxxxx Xxxxxxxx, D., Vandekerckhove, L. and Xxxx, M., 1999. Testing and evaluating the Ephemeral Gully Erosion Model (EGEM) in Southern Europe (SE-Spain and SE-Portugal) and the loess belt (Belgium).

Ephemeral gully erosion studies, possibilities of joint research. USDA-NRCS, National Sedimentation Laboratory - Oxford Mississippi, 23-25 August, 1999.

Nachtergaele, J., Xxxxxx, J., Xxxxxxx, A., Xxxxxx, I., Xxxxxxxxxx, L., Xxxxxx, X., 2000. Ephemeral gully erosion in the Belgian loess belt. International symposium on Gully erosion under Global Change, K.U. Leuven, 16-19 April, 2000.

Xxxxxx, X., Xxxxxxxxxxxx, X., Xxxxxxxxxxx, X., Xxxxxxxx Xxxxxxxx X., Xxxxxxxx, X., 0000.

Gully erosion under environmental change. International symposium on Gully erosion under Global Change, K.U. Leuven, 16-19 April, 2000.

Xxxxxx, X., Xxxxxxxxxxxx, X., Xxxxxxxxxxx, X., Xxxxxxxxxxxxxx, X., Xxxxxxx, X., 0000.

Gully erosion as a missing link in erosion models. COST 623 International workshop on Linkage of Hillslope Erosion to Sediment Transport and Storage in river and Floodplain Systems. Almeria, Spain, 7-11 September, 2000.

Xxxxxxxxxxxx, X., Xxxxxx, X., Xxxxxxxxx, X., Xxxxx, X., 0000. Flow width prediction for concentrated flow on agricultural fields. Cost 623: ‘Snowmelt erosion and related problems’. The Norwegian State Pollution Control Authority (Jordfrosk), Xxxx, Xxxxxx, 00-00 Xxxxx, 0000.

4.3 Organization of the International Symposium

The International Symposium on Gully Erosion under Global Change was held at the K.U. Leuven, April 16-19, 2000. During this symposium the MWISED project was presented through the following papers:

a) Borselli X., Xxxxxxxxxx, S., Xxxxxxxx, P., Xxxxxxxx, V., Xxxxxxxxxxxx, J., Xxxxxx, J., Xxxxx, V., Xxxxx, D. Field experiments for gully initiation. (poster presentation)

b) van de Vlag, D., Jetten V., Nachtergaele J., Xxxxxx J. Event based modelling of gully incision and development in the Belgian loess belt. (oral presentation)

c) Xxxxx, D., Xxxxxxxx, L. Some Further Equation for High-Rate Gully Erosion. (oral presentation)

Besides presentations directly related to the MWISED project, other presentations on topics related to MWISED were discussed.

5. REFERENCES

Xxxxxxx, E.E., Xxxxxxx, M.A., Xxxxx, M.A., Xxxxx, X.X., Xxxxxxx, F.B., Xxxxx, X.C., Xxxxxx, J.M., Xxxxxxxx, J.R. 1995. Soil component. In: Xxxxxxxx, D.C., Xxxxxxx, M.A. (Eds.) USDA – Water Erosion Prediction Project. Hillslope profile and watershed model documentation. NSERL Report n°10, USDA-ARS National Soil Erosion Laboratory, West Lafayette, Indiana: 7.1-7.47.

Bennet, S., Xxxxxx, X., Xxxxxxxx, K.M., Xxxxxx, K.C., 2000. Characteristics of actively eroding ephemeral gullies in an experimental channel. Transactions of the ASAE. 43 (3): 641-649.

Xxxxx, X.X., 1966. Erosion and deposition in the loess mantled Great Plains, Medicine Creek drainage basin, Nebraska. U.S. Geology Survey Professional paper 352(H): 347-350.

Xxxxxx, P.J.J., Xxxxxx, X., Xxxxxx, G., Vandaele, K., 1999. Importance of slope gradient and contributing area for optimal prediction of the initiation and trajectory of ephemeral gullies. Catena, 37: 377-392.

Xxxxxx, G.R., Xxxxxxxx, D.C., Xxxxxxx, M.A., Xxxx, X.X., Xxxxx, L.M., Xxxxxx, S.C. 1995. Hillslope erosion component. In: Xxxxxxxx, D.C., Xxxxxxx, M.A. (Eds.) USDA – Water erosion prediction project. Hillslope profile and watershed model documentation. NSERL Report n°10, USDA-ARS National Soil Erosion Laboratory, West Lafayette, Indiana: 11.1-11.12.

Xxxxxx, X.X., 1980. CREAMS: A field scale model for chemicals, runoff and erosion from agricultural management systems. U.S. Department of Agriculture. Conservation research report 26: 480-485.

Xxxxxx, X.X., Xxxxxxxx, D.E. and Xxxxxx, C.D., 1988. Ephemeral gully erosion model (EGEM). In: Modelling Agricultural, Forest and Rangeland Hydrology, American Society of Agricultural Engineers Publication, 07-88: 315-323.

Nachtergaele J., Xxxxxx X., 1999. Assessment of soil losses by ephemeral gully erosion using high-altitude (stereo) aerial photographs. Earth Surface Processes and Landforms, 24, 693-706.

Nachtergaele X., Xxxxxx J., Xxxxxxxxxxxxxx X., Oostwoud Wijdenes D., Xxxx M., 2001. Testing the Ephemeral Gully Erosion Model (EGEM) for two Mediterranean environments. Earth Surface Processes and Landforms, 26: 17-30.

Xxxx, X.X., Xxxxxxxxx, J.E., 1970. River flow forecasting through conceptual models. Part 1 – A discussion of principles. Journal of Hydrology 10: 282-290.

Xxxxxx, P.C., 1973. Gully erosion in the semi-arid West. MSc. thesis. Colorado State University, Fort Xxxxxxx, CO, pp. 129.

Xxxxxx J.W.A., 1992. Mechanisms of overland-flow generation and sediment production on loamy and sandy soils with and without rock fragments. In: Overland flow hydraulics and erosion mechanics, Xxxxxxx A.J., Xxxxxxxx A.D. (eds), UCL Press, London: 275- 305.

Xxxxxx, X., Xx Xxxx, E., Xxxxxx, X., Xxxxxxxxxxxx, J., Xxxxxx, G., 1999. Concentrated flow erosion rates as affected by rock fragment cover and initial soil moisture content. Catena, 36: 315-329.

Xxxxxxx, X. and Xxxxxxx, X., 1984. A unifying quantitative analysis of soil texture. Soil Science Society of America Journal, 48: 142-147

Xxxxxxx E.T. and Xxxxxxx, R.P., 1961. Critical tractive forces in cohesive soils.

Agricultural Engineering, 42: 26-29.

Vandaele, X., Xxxxxx, J., Xxxxxx, G. and xxx Xxxxxxxx, B., 1996. Geomorphic threshold conditions for ephemeral gully incision. Geomorphology, 16(2): 161-173.

Vandekerckhove, X., Xxxxxx, J., Xxxxxxxx Xxxxxxxx, D., xx Xxxxxxxxxx, T., 1998. Topographical thresholds for ephemeral gully initiation in intensively cultivated areas of the Mediterranean. Catena, 33: 271-292.

Xxxxxxxxxx X.X., Xxxxx X.X., 0000. Predicting rainfall erosion losses – A guide to conservation planning. United States Department of Agriculture Agricultural Handbook, 537.

Xxxxxxxx, XX. 1999. Method to predict cropland ephemeral gully erosion. Catena 37: 393-399.

XXXXXXXXX UNIVERSITY (CRANFIELD)

REPORTING PERIOD: 1 APRIL 1998– 30 JUNE 2001

Contractor:

Cranfield University

Responsible scientist: Professor X X X Xxxxxx

Address: Xxxxxxxxx Xxxxxxxxxx xx Xxxxxx Xxxxxx, Xxxxxxx XX00 0XX Xxxxxx Xxxxxxx

Telephone: + 44 - (0)1525 - 863059

Fax: + 44 - (0)1525 - 863344

E-mail: x.xxxxxx@xxxxxxxxx.xx.xx

Other scientists: Dr X. X. Xxxxxxx, Dr. X. Xxxxx, X. Xxxxxx

Scientific activity

1 WP1. Within storm changes in infiltration

ST2 Soil Roughness

1.1 Assessment of surface roughness and depression storage

The objective of the research on surface roughness and depression storage was to see whether a better procedure could be developed to assess the volume of depression storage than the method currently proposed in the EUROSEM User Manual (Morgan et al. 1998). The present approach uses measurements of roughness made with a 1-m long chain with 5 mm links along transects in the field to obtain a roughness index (RFR), defined as the ratio of straight-line distance of the transect (X) to the actual distance measured over the microtopographic irregularities (Y):

A formula developed by Xxxxxxxxx is then applied to estimate the depth of depression storage from the roughness index:

This study investigated the feasibility of estimating depression storage from digital terrain models (DTMs) of the surface, validating the method against laboratory measurements. The work was undertaken in three phases:

• establishment of reliable data on depression storage for surfaces of varying roughness in the laboratory;

• development of a methodology for estimating depression storage from a DTM; and

• validation of the method by comparing estimates with the laboratory values.

1.1.1 Laboratory determination of surface roughness

Samples of a xxxxx xxxx soil of the Cottenham Series, a locally-available erodible soil, were air dried and passed through a 3-mm sieve. The clods remaining on the sieve were sprayed with diluted PVA glue (ratio 1:3) and then placed in an oven at 110°C until dry. The dry clods were placed into wet cement in three metal boxes, measuring 2 x 1 m, to give three soil surfaces of low (< 4 mm), medium (> 4mm) and high (> 4 mm with added topography to give a maximum height difference of 100 mm) micro-topographic roughness.

In order to ensure the stability of each surface, protect it from raindrop impact and avoid infiltration, the surfaces were covered with PVC-based cling film which was glued to the soil with PVA glue. This allowed each surface to be used again as replicate treatments. The boxes were placed in a runoff rig to which simulated rainfall was applied at nominal intensity of 103 mm/h. Measurements were made of the time for runoff to start, the time at which a constant runoff rate was achieved and the roughness of the surface using the RFR index of EUROSEM. Since there was no infiltration, the accumulated rainfall at the onset of overland flow was considered to equal the surface depression storage. Four different slopes, 1.7°, 3.7°, 5.7° and 6.9°, were used with three replications of each slope-roughness level combination.

The results of the experiments showed that depression storage decreased with slope, the form of the relationship depending on the surface roughness (Figure 1). With high roughness, depression storage declined much faster compared to medium and low roughness. Although graphical plots indicate that the relationships might be curvilinear, linear relationships provide an excellent description of the trends (Table 1). These findings correspond to those of Xxxxxx (1984) and Xxxxxx et al. (1988) who obtained linear relationships between depression storage and slope at constant roughness, using digitised surfaces. However, Xxxxx and Xxxxxxxx (1990) found an exponentially decreasing relationship using Markov-Gaussian simulated surfaces and Xxxxxxx (1991) obtained a linear relationship when using a cube model and an exponential relationship with an inverted pyramid model. The findings indicate that the relationship between slope and surface roughness depends on the type of roughness, e.g. clod or micro-topographic roughness.

Figure 1 Relationship between depression storage and slope for three surfaces of different roughness

Table 1. Relationships between depression storage (D; mm) and slope (S; degrees) for the surface tested in the laboratory at Cranfield University

Surface RFR (cm/m) Linear regression R2

1	9.70	D = 00.0000 - 0.0000 S	0.9939
2	8.01	D = 0.0000 - 0.0000 S	0.9399
3	3.39	D = 0.0000 - 0.0000 S	0.9414

The measured levels of depression storage are very high compared with those predicted by Xxxxxxxxx’x formula (Table 2). They are, however, compatible with measurements made by Xxxxx and Xxxxxx (1979) and Xxxxx (1996) and estimations made by Xxxxxx et al. (1999) which generally range between 1 and 12 mm, depending on the roughness of the surface. In contrast, Helming et al. (1993) give values between 0.06 and 0.6 mm for surfaces of similar degrees of roughness. Xxxxx and Xxxxxxxxx (1979) and Xxxxx and Xxxxxxxx (1990) also produce estimates of depression storage which are less than 2 mm.

Table 2. Comparisons of depression storage measured in the laboratory on a 1.7° slope with those estimated from Auerswald’s formula

Surface RFR (cm/m) Measured depression

storage (mm)

Estimated depression storage (mm)

1 9.70 13.00 0.0176

2 8.01 3.20 0.0111

3 3.39 2.87 0.0032

1.1.2 Establishment of methodology for estimating depression storage from DTMs.

Images for developing the methodology were obtained from two sources:

• the laboratory experiments carried out at Cranfield University, giving the three surfaces described above; and

• laboratory investigations undertaken at the Istituto per la Genesi e la Ecologia del Suolo, Firenze, giving a further two surfaces.

Pairs of stereophotographs were taken of the surfaces ensuring that the portion of the soil surface for detailed analysis was contained completely in the area of overlap between adjacent photographs. At Cranfield University photographs were taken from a height of 280 cm using a pair of digital KODAK camera (type DCS 420 with a Nikkor lens of 18.392 mm), placed side-by-side. At Firenze, the

photographs were taken with a single Rollei 35 Metric Camera with a focal length of 40 mm and a calibrated graticule. To obtain the stereophotographs, the camera was physically moved between exposures.

Photogrammetry

The negatives of the photographs taken in Firenze were scanned as grey-scale images at 1200 dpi to obtain TIFF files suitable for use in the digital photogrammetry software. The digital images from the Kodak DCS420 were transferred from the cameras and converted to TIFF files. Calibration of the Rollei camera was obtained from the calibration certificate supplied with the camera. As the focus of this study was to examine the utility of the method for determining surface roughness rather than testing the photogrammetric fidelity of the camera, calibration for the Kodak DCS420 cameras was obtained from Warner and Xxxxxxxxxx (1997). Image registration, correction and processing were carried out following standard photogrammetric procedures using the block tool of the ERDAS Imagine OrthoMax software running on a UNIX platform. After triangulation, the stereopairs were used to create a 10-mm resolution DTM of each surface. Stereo editing of the Silsoe surface DTMs was required to ensure that all the heights of the elevation points within the soil box corresponded with the stereo perception of the height of the soil surface. This was necessary because the super wide-angle lens used on the Kodak cameras caused the height points to fold back upon themselves towards the image edge. An ortho-rectified image of each soil surface was then created from the corrected DTMs.

ARC/INFO processing

The DTM and ortho images were exported from ERDAS Imagine as ARC/INFO grids (raster files) with elevation values forming floating point values for each pixel in the DYM grid. Prior to the calculation of depression storage, a polygon coverage was digitised from the ortho image of each soil surface. This was used to define the extent of the DTM to process. In the case of the Cranfield surfaces this corresponded to xxx xxxx of the soil box. The Italian soil surfaces had a defined rectangle within the soil box. The ARCINFO GRIDCLIP function was used with the respective polygon coverages to cut the areas of interest out of the DTM images.

The ‘clipped’ elevation grids were then processed using the ARC/INFO ARC Macro Language (AML) to compute the depression storage. For this purpose, it was necessary to have a clear definition of ‘depression storage’. Depression storage was therefore defined as ‘the sum of the volumes of water held in depressions within the defined watersheds, divided by the sum of the plan areas of the watersheds’. The AML calculated the depression storage according to this definition in each individual watershed within the DTM. No attempt was made to model flow from one watershed to another. Volume was reported in m3 and area in m2 and the resulting depth in mm.

AML programme: description of process

Prior to running the AML, an elevation point file (%.WS_PT%) is prepared from the

elevation raster derived from the OrthoMax software. This is used as input to the AML to construct a Triangulated Irregular Network (TIN) for each watershed. The main processes followed by the AML are described in Table 3.

The main components of each watershed are (Figure 2):

• a sink (the lowest point within the watershed)

• a watershed boundary (the line of highest points from which water will drain towards the sink)

• a pour point (the lowest point along the watershed boundary over which water will spill if the sink is filled to capacity).

Figure 2 Grid showing elevations and designations for watershed definition

Results

The estimated values of depression storage for the five surfaces, based on the DTMs, are given in Table 4.

Table 3. Steps in the AML procedure to determine depression storage

S Procedure

e p

1 On running AML, the user is prompted to enter the name of the ‘clipped’ elevation grid and a short ID for the data sets that will be created.

2 AML creates a point coverage from the elevation grid consisting of a point for every pixel in the original grids.

3 Using the hydrological commands available within the GRID module of ARC/INFO, the number and extent of watersheds in the study area are determined and defined using the following procedure:

(a) determination of the flow direction for each pixel in the elevation grid;

(b) identification of the sinks in the study area from the flow directions;

4 The resulting watershed grid is converted to a ‘coverage’ in order to determine the surface area of each defined watershed.

5 The watersheds are numbered and the total number of watersheds determined. The user is then asked to enter the start and finish numbers of the watersheds for further processing; this feature enables the

user to specify a subset of watersheds for analysis rather than processing the whole file.

6 The AML cycles through the selected set of watersheds, in turn, following this sequence:

. (a) the first watershed is selected and named (%.DATASET_ID%_%.WS%)

(b) using the GRIDPOLY command, a coverage is created of the watershed

(%.DATASET_ID%_%.WS%_cov1); this is used later to cut out the elevation data from the elevation grid in order to calculate the storage volume for the watershed

(d) using the watershed ‘coverage’ to define the boundary, GRIDCLIP is used to cut out or clip the elevation pixels within the watershed for further analysis

(e) the individual watershed area of elevation pixels is used to select all points with elevations less than or equal to the elevation of the pour point

(f) if there are no points that meet the criterion for (e), i.e. there is no area within the watershed that lies below the pour point, a value of -999999 is assigned to the variable %wstest2% and the program calls a subroutine (loop_end) which jumps over the remaining steps and moves on to the next

watershed

(g) if the variable %wstest2% is assigned a value other than -999999, the program creates an integer grid of the watershed elevation values; values are rounded down if the lie between 0.1 and 0.49 and rounded up if they lie between 0.5 and 0.99

(h) a coverage (%.DATASET_ID%_&.WS%_poly2) of the integer grid is created using the GRIDPOLY command

(i) the watershed elevation coverage is required to perform a HARDCLIP when creating the Triangulated Irregular Network (TIN) of the watershed; to enable this step, a new field (height) is added to the coverage polygon attribute table and all records are assigned the value -9999

(j) a TIN of the area bounded by the watershed is constructed from the elevation point file (%.WS_PT%) prepared prior to the running of this AML from the elevation raster derived from the OrthoMax software; the hardclip coverage is set to %.DATASET_ID%_%.WS%_poly2

(k) the volume of the TIN below the pour point elevation is calculated and stored in an attribute table within INFO (%.DATASET_ID%_dep.att)(l) unwanted TINs, coverages and grids are deleted before returning to the loop to analyse the next watershed

7 The AML cycles through all the watersheds and a second AML is executed to sum the volumes of the individual watersheds

Partner’s report: CRANFIELD

Table 4. Results from the ARC/INFO depression storage calculations

Surface	Number of watersheds	Area of watersheds (m2)	Volume of depression storage	Depth of depression storage
			(m3)	(mm)
Cranfield (high roughness)	524	1.98	0.0006600	0.33
Cranfield (moderate roughness)	693	1.82	0.000656	0.36
Cranfield (low roughness)	514	1.76	0.000241	0.14
Firenze 1	511	0.0038	0.000013	3.42
Firenze 2	126	0.0035	0.000003	0.86

1.2.5 Discussion

The automated generation of height values within OrthoMax was successful. An important aspect of this method to consider is the time required to derive the data for the calculation of depression storage. Data capture is instantaneous once the cameras are in position. Triangulation, DTM and orthoimage creation are rapid due to the small file sizes involved. However, all surfaces had to be carefully checked to confirm that the automated heighting of DTM points was correct with respect to the stereo perception of the soil surfaces. Some heights were incorrect in the Silsoe surfaces due to xxx xxxx problems described earlier. This adds to the processing time of the data. On average, for the rougher surfaces at Cranfield, the time required to derive the clipped DTM was 1.5 days. The time required for the calculation of depression storage varied according to the number of watersheds to process. As an example, the rough Cranfield surface took 5 hours to process using the ARC/INFO AML.

Close examination of the processed data files shows that many watersheds defined by the AML do not have any depression storage associated with them and consequently return a volume of 0.00000 m3.

This is in part due to the default field width of five decimal places applied by ARC/INFO in the data table created by the VOLUME command. It is possible to create a volume measurement for some of these shallow depressions by multiplying the x, y and z dimensions by an appropriate factor, for example, 10. This was done for the rough Cranfield surface giving a volume of 0.00067 m3.

In a physical model there is likely to be some residual volume of water retained in all watersheds due to surface tension and micro relief that is not modelled by the AML. The AML assumes no retention of water other than that held within the watershed below the pour point of the depression.

The resolution at which the surface is modelled will have a significant effect on the output. The computed volume is influenced by the number of points used in the TIN procedure, which in turn is influenced by the resolution of the DTM. The higher the resolution of the DTM, the closer the modelled surface will be to the actual surface, and visa versa. A horizontal resolution of 10mm was

used in this study because this was seen to be a good compromise between the amount of data to be processed and how representative the DTM was of the soil surface.

The AML could be modified to incorporate other features. The effect of tilting the surface, variations in rainfall intensity and soil permeability could be useful extensions to the present model.

1.2 Discussion

Values of depression storage derived from the DTMs are very low compared with those measured in the laboratory and the estimates from Xxxxxxxxx’x formula. There is no reason to believe that any of the values are incorrect, so the differences must relate to the way in which they have been derived and, therefore, to the definitions used for depression storage.

Depression storage can be considered as the amount of water held in the surface depressions, none of which runs off, and which may subsequently be evaporated or infiltrated. It excludes surface detention which is storage of overland flow in transit (Chorley, 1980). In the laboratory experiments described above, depression storage is defined as the amount of water held in surface depressions when runoff starts to flow from the base of the slope. By the time this occurs, water will have already filled some of the smaller depressions and overflowed into larger ones, contributing to the storage there until their pour point is reached. The measurements made from the DTMs give the storage of each depression and do not allow for water moving on the surface from one depression to another but not flowing to the base of the slope. The latter is similar to the methodology adopted by Xxxxxxx et al. (1993) who considered only the storage capacity of each depression from estimates of its circumference, area and volume. Xxxxxxxxx’x formula is based on data from Helming (1992).

The key questions for the estimation of depression storage are to decide when runoff begins and how best to describe the dynamics of storage and runoff generation. The traditional hydrological approach (Xxxxxx, 1937) is to assume that depression storage must be satisfied before runoff can begin. In reality, however, surface depressions are not all the same size and volume and flow from the smaller depressions can begin before the storage of the large ones is filled. In the laboratory it is possible to decide that runoff occurs when flow reaches the bottom of the slope. In the field, the slope length may be tens of metres or more and runoff can produce serious erosion before it reaches the base of the slope. Either a threshold length of flow has to be determined which, when exceeded, defines the point at which runoff starts or the approach of Helming et al. (1993) has to be adopted which means that runoff occurs as soon as any of the depressions overflow. If the latter approach is taken, the situation arises in which runoff is occurring whilst depression storage is still increasing. The question then arises as to whether it is necessary to model the dynamics of this phase of runoff generation.

Over a complete storm it is clear that the rainfall which goes to depression storage does not contribute to runoff and that, numerically, Xxxxxx’x (1937) approach holds. Since the volumes of surface depressions are rather small, generally less than an equivalent depth of 0.5 mm (Helming et al. 1993), it could be argued that for the storms in which most erosion occurs there is no need to model depression storage at all. However, if EUROSEM is to be used for storms of high frequency and moderate

magnitude, which will be the case for pollution modelling (Xxxxxxx et al. 2001), there is a need to retain the depression storage component which, on rough (ploughed) surfaces would be sufficient to store

20-30 per cent of the rainfall.

DTMs with a 10-mm resolution can be used to obtain reliable estimates of depression storage but the method is rather time-consuming, particularly for rough surfaces. The present method of measuring roughness in the field and estimating depression storage from a roughness index is an appropriate way of obtaining data quickly with a sufficient level of accuracy for modelling purposes, given the small values involved. Except for very high values of roughness (25-40 cm/m), Xxxxxxxxx’x formula seems to underestimate depression storage by up to an order of magnitude.

1.3 Conclusions and recommendations

For modelling purposes, depression storage is best defined in relation to the volume of all the depressions on the surface to the point at which each depression overflows. This avoids the problem of having to define when runoff occurs. Stereophotographic pairs of the slope surface can be used to obtain DTMs from which depression storage can be calculated using ARC/INFO ARC Macro Language. However, the method is time consuming. The present procedures for estimating depression storage from field measurements of roughness are adequate as data input to EUROSEM provided that improvements can be made to Auerswald’s formula. Further work is recommended to obtain data on depression storage from the literature and develop a new algorithm for estimating depression storage as a function of roughness.

2 WP4: Simulation of within storm erosion dynamics.

2.1 Recoding of EUROSEM

The focus of our work since the last interim report has been on the completion of the graphical user interface (GUI) and the incorporation of new routines for gully erosion and infiltration.

It was originally proposed to recode EUROSEM in Fortran 90. However, after contracting a programmer we were advised to use an object orientated language. To facilitate the development of a Graphical User Interface (GUI) the Delphi language was chosen.

The EUROSEM model has been completely recoded from Fortran 77 into Delphi. The resulting product has received the name EUROSEM 4 Win. The Window based application includes a graphical interface which incorporates sheets for catchment definition (Figure 3), rainfall description (Figure4) and the display of output (Figures 5 and 6). The model can be parameterised on an element- by-element basis (Figure 7) or use can be made of data dictionaries, which facilitate the use of information derived from pedoalgorithms or other environmental data sources. The software also incorporates full help support and online documentation.

The major components of the application and some details of their implementation are as follows (Figure 8).

I. Data declaration section. This describes the complete set of global objects, data types and structures. It provides the visibility of data for all subroutines involved in the data processing.

The main data structures are:

- Simulation record, which keeps the information on the simulation options and the data sets used in the simulation.

- Catchment record, which keeps the general catchment parameters and the list of elements comprising the catchment.

- Element data record which incorporates the data on each element along with data describing its appearance and methods to realise its visualisation. Element numeric data are used by simulation routines; visualisation data and methods are used by GUI.

- Rainfall data are declared as static array of records and used by simulation routines to perform the simulation.

- The observed data record is used for displaying measured data sets to compare them with modelled results.

- Output data are declared as a several sets of records with different structures (dynamic output, static element-wise output, static summary output).

- Data dictionary structures.

- Runtime messages and help references which are sensitive to context.

II. The GUI is the main controlling component, maintaining interaction with the user, as well as interaction between the other components of the software.

Interaction with the user is realised by a set of controls within subroutines that implement the response to all user actions and perform the required operation.

The other important function of GUI is to provide access to a visualised input data to make data entry and editing possible.

During the simulation run, the GUI accepts the information on the simulation progress generated by the simulation routines and displays this information on a dedicated panel. At the end of simulation run, the GUI displays output data in a dynamically created output form from the temporary output disk files created by the simulation routines.

III. The data management section is a set of routines implementing data saving and loading to and from the disk files having binary internal structure. To provide additional data safety, before file saving, a backup copy of the previous version of each file is created.

IV. The simulation algorithms are the core of the EUROSEM model. Logical and numerical interpretation of the model was translated from the Fortran 77 code based on the latest EUROSEM version 3.2 of 11/97. In the Windows based version simulation routines are declared as the methods of a thread object. The simulation thread is executed as a separate process, constantly generating messages to notify the system on its progress and status. The messages are processed by the main application to display the simulation progress. At the end of a successful completion of the simulation, a set of temporary output files is created, which are used by the output display functions of the GUI. If the simulation is terminated, the simulation thread notifies the calling application about the reason for termination (i.e. cancellation by user or abortion due to the run time errors). The detailed simulation progress is registered in the log file, which can be used later for debugging purposes or to trace the sequence of calculation.

If necessary, the simulation unit can be compiled and distributed as a dynamically linked library, which allows users to embed EUROSEM routines into other specialised software.

Using the thread object for simulating routines has the advantage of allowing several simulation processes to run simultaneously, if required. In this case, using a thread provides an effective use of system time and faster execution of the simulation (e.g. on multiple rainfall simulations).

V. A Data Conversion Utility was developed to help DOS version users convert their data into the format compatible with the Windows based version. It is accessible from the main menu of the EUROSEM 4 Win.

VI. The Data Dictionaries Utility (DDU) is one of the additional features designed to simplify the input of element data. It also may be used for user data systematisation. For the DDU the creation of the data dictionaries, data input and editing, is implemented as a stand-alone application, accessible from the main menu of the EUROSEM software. The format of data files (soil, surface condition, and vegetation) created by the DDU is recognisable by EUROSEM 4 Win. Using an option of the data entry form, the data from the dictionary can be inserted into the corresponding fields of the data entry form. Both EUROSEM 4 Win and the Data Dictionary Utility share a common dynamically linked library implementing all functions dealing with the data dictionaries.

VII. Online user help is developed in WinHelp format. Access to the help content is available from the main menu featuring standard WinHelp keyword search and global search functions. Some elements of user interface provide direct links to help topics. Also, context-sensitive help is available from the data entry form, giving the user a description of the input parameters.

The main advantages of the Windows based version over the DOS based one are:

- convenient and easy to use interface

- visualisation of input and output data

- export of output data into graphic and table data formats

- using of time saving options on the data editing, like element copying, data paste and copy via Windows clipboard, data export from the data dictionaries, etc.

- removal of DOS version bugs which were detected and eliminated in the Windows-based version.

In addition to the basic functions available in the DOS version of EUROSEM, the Windows based version has the following new features:

- multiple rainfall simulation

- incorporation of new algorithms of infiltration and ephemeral gully formation developed within the framework of the MWISED project.

Testing and debugging. Currently the main routines have been programmed, determining application functionality, its appearance and user interface concept. Considering the relatively short time devoted to the application development, the authors have not had an opportunity to test it comprehensively.

Although the application has been constantly tested during its development, using a sample data set, the testing is not complete and the application cannot be guaranteed bug free. In addition to the further debugging of the application by its developers, feedback with the potential users on bug reports is strongly desirable.

2.2 Incorporation of new infiltration routines

New routines, using equations supplied by CNR Firenze, have been added to EUROSEM to describe within-storm changes in infiltration.

Algorithmically, these equations have been embedded into existing EUROSEM subroutines for calculating the infiltration rate

User interface provides controls to select between infiltration equations Xxxxxxxx-Xxxxx and Xxxxx- Seytoux.

The dynamics of the infiltration rate calculation is illustrated by the simulation results for the sample data set having the parameters given in Figure 9. Figures 10 and 11 display the infiltration rate for each calculated time increment at each distance node of the investigated plane using both algorithms. Figure 12 shows the comparative dynamics of the infiltration at the last node of the plane.

Compared with results from the Xxxxxxxx-Xxxxx equation, the Xxxxx- Seytoux equation gives longer predicted time to saturation; the infiltration decreases more slowly while a rainfall continues, and faster after the rainfall recession. Also, the global volume balance error is less when using Xxxxx- Seytoux equation.

More informative assessment of the reliability of both algorithms can be done through comparison of the modelled and measured data.

Figure 3. Catchment definition screen showing plane elements (rectangles) and channel elements (arrows).

Figure 4. Rainfall input dialogue box.

Figure 5. Output displayed as text.

Figure 6. Output displayed graphically.

Figure 7. Parameterisation dialogue.

I. Data declaration:

simulation data record

catchment data record

element object

rainfall data record

III. Data management functions:

- save

V. Data Conversion Utility

IV. Simulation thread methods:

- simulation run

- messaging

- saving simulation results

II. GUI functions:

access to application functions

- response to user action

- input and output data visualisation

- interaction between other components of application

- access to the context-sensitive help and to help topics

Vi. Data Dictionaries DLL:

- displaying data dictionary record

- filling in the data entry fields with

VII. Online help:

- help contents

- keyword search

Data Dictionaries Utility (stand alone) :

- creation of the data dictionaries;

Partner’s report: CRANFIELD 17

Figure 8. Diagrammatic representation of program fucntioning.

CATCHMENT PARAMETERS

Total number of elements: 1 Overland flow length : 100 Total simulation time : 60 Time step: 0.5

Air temperature: 10°C

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

PLANE 1, INPUT DATA

Element geometry:

Length: 50.00 m Width : 20.00 m

Soil hydrology

Saturated hydraulic conductivity: 50.00 mm Capillary drive: 240.00 m Porosity: 0.45 % Moisture Initial: 0.40 Maximum: 0.42 % v/v

Soil physical properties

D50: 125.00 µm3 Specific gravity: 2.65 mg/m Erodibility: 1.60 g/J Cohesion: 5.00 kPa

Depth to non-erodible layer : 3.00 m Soil particle detachment exponent: 2.00

Rocks

Rock fragment cover : 0.40% Rock fragment content : 0.40% v/v

Rock fragments are not embedded into a surface

Surface conditions Number of rills: 0.00

Rill width: 0.00 m Rill depth: 0.00 m Rill side slope: 0.00 Slope: Rill: 0.00 Interril: 0.10

Xxxxxxx'x N: Rill: 0.00 Interrill: 0.06

Partner’s report: CRANFIELD 19

Roughness: 1.00 cm/mm Recession factor: 10.00 cm Rills dimensions are constand along the slope

Vegetation parameters

Vegetation cover: 0.00 Plant height: 0.00

r’s report: CRANFIELD

Infiltration, mm/h (by Xxxxxxxx-Xxxxx)

Node 1

Node 3

Node 5

Node 7

Node 6

Node 5

Node 4

Node 3

Node 2

Node 1

(Global volume balance error =0.0635%)

Figure 10. Infiltration rate (mm/h) plotted against time at different computational nodes for the Xxxxxxxx-Xxxxx equation.

Infiltration, mm/h (byMorel-Seytoux)

Node 1

Node 3

Node 5

Node 7

(Global volume balance error =0.0451%)

Figure 11. Infiltration rate (mm/h) plotted against time at different computational nodes for theMorel -Seytoux equation.

Node 7

Node 6

Node 5

Node 4

Node 3

Node 2

Node 1

0.5

6.5

12.5

18.5

24.5

0.5

6.5

12.5

18.5

24.5

30.5

36.5

42.5

30.5

36.5

42.5

48.5

54.5

Partne 20

0.000018

0.000016

0.000014

0.000012

0.00001

0.000008

0.000006

0.000004

0.000002

Comparison of the infiltration rate dynamic (node 7)

R,mm/x

Xxxxx-Seytoux

Xxxxxxxx-Xxxxx

Figure 12. Comparison of infiltration rate using the Xxxxx-Seytoux and ParlangeSmith equations

0.5

11.5

22.5

33.5

44.5

55.5

2.3

Incorporation of Gully Routines

In the process of the application development an attempt has been made to incorporate ephemeral gully calculations into the EUROSEM model.

The main idea on gully routines implementation implies in declaring of the new type of element called thalweg (valley bottom) where the initiation of ephemeral gully is possible under certain rainfall conditions. The new element is very similar to the plane element having one rill along the main slope in the middle of it. The shape of the rill is described as a trapezoid having the same side slope as the interrill slope of the thalweg element, thus representing a continuous surface of the specified shape. Like the plane, the thalweg element can accept runoff and sediment from the upslope element. In addition, it can receive runoff contributed by right and left hand planes (Figure 13). The runoff from the upslope plane is added to the first (topmost) distance node of the thalweg element; the contribution from side planes is added to a first node of an interrill area.

Figure 13. Ephemeral Gully element with contribution, left, right and upper planes.

At the end of each time step a gully initiation threshold is checked. As a threshold of gully initiation and initial gully parameters the following relationship has been used (provided by MWISED partners Xxxx Xxxxxx and Xxxxxx Xxxxxxxxxxxx): the threshold based on the discharge Q for gullies in cultivated European fields is 0.002 m3 s-1 < Q < 1 m3 s-1 and the initial width W of a gully is W = 2.51 Q0.41.

While the value of discharge is below the threshold value, the normal EUROSEM routines implementing rill and interrill calculations apply. When the gully threshold is exceeded, the rill dimensions and shape is recalculated: the width of rill is determined as described above, the cross section of gully is determined as rectangular, and the depth of the gully is calculated so that accommodate the available runoff. The width of interrill area is recalculated, and the supply of sediment is computed from the difference in the cross section area before and after gully formation. Until the

discharge falls below the threshold the gully routines are applied, when discharge falls below the threshold value, the algorithm switches back to standard EUROSEM rill and interrill routines.

The modelled gully parameters are stored into the special data structure, which could be used for analysis and visualisation.

The approach described is illustrated on the fragment of the Rottingdean catchment. The catchment configuration is shown in figures 14.

Figure 14. Diagrammatic representation of the upper reaches of the Rottingdean catchment.

Figure 15 shows the hydrograph for a test simulation compared to the critical threshold. Figure 16 shows the dynamical change of the modelled gully cross section (a- at the beginning of simulation; b – the time step before gully initiation; c – gully initiation; d – gully growth; e – gully cross section when the discharge decreases; f – final gully cross section).

0.45

0.4

0.35

0.3

0.25

0.2

0.15

0.1

0.05

-0.05

Discharge

CriticalDischarge

Rainfall

2.5

1.5

0.5

0.45

0.4

0.35

0.3

0.25

0.2

0.15

0.1

0.05

FlowVel GullyWidth GullyDepth

Head

Discharge

101

111

Figure 15 Hydrograph, hyetograph and threshold for gully initiation for the Rottingdean catchment (Discharge is plotted on the left hand axis, Rainfall on the right hand axis).

100

109

118

Figure 16, Changes in gully dimensions through time for the Rottingdean test catchment (Gully dimensions and flow velocity are plotted on the left hand axis, discharge is plotted on the right hand axis).

This approach has certain limitations. Since the main functions of the thalweg element implementation have been adopted from the EUROSEM plane element, it is not possible to define different parameters for the different sides of the thalweg element, such as slope, width, and vegetation. Also, it is impossible to describe a rilled surface onthe side of the thalweg. This can be overcame by the defining the thalweg element as narrow as possible, so that the conditions at both sides were similar, and describing differences in conditions of the contributing areas by parameterisation of the contributing side planes.

The advantage of this approach is that it could be implemented using other parameters (or combinations of parameters) as a thresholds and for initial gully dimensions calculations, namely, runoff cross section area, cross section dimensions, flow velocity, slope.

The development of ephemeral gully routines is currently at the experimental stage. Before introducing it to the user, more tests and trials should be made, the visualisation functions and user interface service functions for the thalweg element will need to bebe programmed.

3 ST2 field validation

Late completion of the finalised EUROSEM model has prevented us from testing the model against the Woburn dataset within the timespan of the project. However, it is our intention to fully test the model over the coming months.

4 References

Xxxxxxx, X.X. 1980. The hillslope hydrological cycle. In Kirkby, M.J. (ed), Hillslope hydrology. Wiley, Chichester: 1-42

Xxxxxxx, X. 1991. The influence of soil surface configuration on depression storage and soil loss. MSc Thesis, Xxxxxxxxx University.

Xxxxx, X. 1996. Soil surface roughness an infiltration in the savanna ecosystem and its impact on erosion. Geografisk Tidsskrift 96: 32-39.

Xxxxxx, X., Xxxxxxxxxx, P. and Xxxxxxxx, E. 1999. Roughness indices for estimation of depression storage capacity of tilled soil surfaces. Soil & Tillage Research 52: 103-111.

Xxxxxxx, X. 1992. Die Bedeutung des Mikroreliefs für die Regentropfenerosion. Bodenökologie und Bodengenese 7, 155 pp.

Xxxxxxx, X., Xxxx, Xx.X., Xxxx, X. and Xxxxxxx, H. 1993. Characterization of rainfall - microrelief interactions with runoff using parameters derived from Digital Elevation Models (DEMs). Soil Technology 6: 273-286.

Xxxxxx, X.X. 1937. Hydrologic interactions of water and soils. Proceedings of the Soil Science Society of America 1: 401-429.

Xxxxx, X. and Xxxxxxxx, J.M. 1990. Depressional storage for Markov-Gaussian surfaces. Water Resources Research 26 (9): 2235-2242.

Xxxxxx, D.L., xxx Xxxxx, D.M. and Xxxxxxxx, R. 1988. A model of the effects of tillage-induced soil surface roughness on erosion. 11th International Conference of the International Soil Tillage Research Organisation, Edinburgh, 11-15 July 1988, Volume 1: 373-378.

Xxxxx, I.D. and Xxxxxx, C.L. 1979. Estimating microrelief surface storage from point data. Transactions of the American Society of Agricultural Engineers 22: 1073-1077.

Xxxxxx, R.P.C., Xxxxxxx, J.N., Xxxxx, R.E., Xxxxxx, X., Xxxxxx, J.W.A., Xxxxxxxxx, K., Chisci, G., Xxxxx, D., Xxxxxxx, M. and Xxxxx, A.J.V. 1998. The European Soil Erosion Model (EUROSEM): Documentation and User Guide. Xxxxxxxxx University at Silsoe

Xxxxxx, X.X. 1984. Depressional storage on tilled soil surfaces. Transactions of the American Society of Agricultural Engineers 27 (3): 729-732.

Xxxxxxx, X.X., Xxxx, J.A. and Xxxx, T.M. 2001. The selective removal of phosphorus from soil: is event size important? Journal of Environmental Quality 30: 538-545.

Xxxxx, X. and Xxxxxxxxx, W.T. 1979. Quantitative description of depression storage using a digital surface model. Journal of Hydrology 42: 63-90.

Xxxxxx, X. X. & Xxxxxxxxxx, B. R. 1997. Multiplotting with images from a Kodak DCS420 digital camera. Photogrammetric Record 15(89): 665-672.

5 Appendix 1

Introducing the Xxxxx-Xxxxxxx & Xxxxxx (1981) equation as infiltration routine

Xxxxxxx Xxxxxxxx CNR-IGES Florence , IT

xxxxxxxx@xxxx.xx.xxx.xx

(static model)

Basic Soil parameters:

G = net capillary drive or wtting front suction as in Mein and Xxxxxx /Green-Ampt models (in mm)

Ks = saturated conductivity (in mm/h)

The composite parameter B is given as:

B = G(qs

−qi )

(like in eurosem)

where

(qs

− qi )

is the initial water deficit with respect the natural saturation

Document Metadata

Table of Contents

MODELLING WITHIN-STORM

Document Metadata