Climate models. Downscaling
Climate models. Climate models are numerical representations of the climate system and are based on physical properties and feedback processes. Coupled atmosphere/ocean/sea-ice general circulation models, commonly referred to as global climate models (GCMs) provide a comprehensive representation of the global climate system. This modelling has been conducted through a series of Coupled Model Intercomparison Projects (CMIP), the latest of which is CMIP5. However, these models provide outputs at a high aggregation level: the horizontal resolution of the GCMs involved in CMIP5 was between 100 and 300 km. Therefore, to derive a finer resolution at local-scale , different downscaling approaches are used. Dynamical downscaling uses the output of GCMs to force regional climate models (RCMs) to obtain a finer representation of climate conditions, producing results in the order of 10 to 40 km resolution. The Coordinated Regional Climate Downscaling Experiment (CORDEX) and the EUROCORDEX database provides the most recent and highest resolution simulations for Europe, covering the historical period and different future scenarios with different RCMs. However, the various global and regional models have different characteristics and this means that even for a single SSP and RCP scenario, there will be a large range of projected change from different climate models, which in turn will affect the level of economic costs. As an example, for the European domain, the differences in climate models (even for downscaled EUROCORDEX data) are large, as shown by Xxxxxxx et al. (2014). For some parameters these changes are robust in direction but involve a wide range (e.g. the exact level of temperature warming), while for others, the range is wide and can even vary in terms of sign (e.g. rainfall in some parts of Europe). In impact assessment, an ensemble of model runs is typically used (a group of parallel model simulations for the same RCP) with analysis of both the average and ensemble range. In many cases, for impact analysis, a number of global or European climate model runs are used that reflect drier or wetter, or hotter or cooler models. This sampling approach has been used in previous European research projects such as IMPACT2C and IMPRESSIONS, which both used regional climate model sampling (including both global driving models and regional climate models) for multiple RCP scenarios. However, the inclusion of climate model uncertainty expands the matrix above along a third dimension, a...
Climate models. Climate models are essentially combinations of mathematical equations that represent different nature processes in the climate system. These processes include radiation on the earth surface, cloud physics, atmospheric and oceanic circulation, chemical cycles, growth of vegetation, etc. Atmospheric models with different resolution have different representation of these aforementioned processes, which in principle aim to reduce the complexity of the computation while ensuring the accurate representation. Initially climate models are built to study the physics of the nature and later on they have been used to generate projections showing consequences of different scenarios, for example, different CO2 concentration or radiation scheme which are likely the consequences of public policy making. Climate models are capable to predict weather only few days ahead of time, but their ability to make reasonable predictions of statistics of weather, i.e. climate prediction, is retained. Thus, climate prediction involves running climate models at least for several seasons and commonly for several years. Downscaling is a method for obtaining high resolution data from relatively coarse resolution global climate data. Typically, downscaling involves statistical downscaling or dynamical downscaling. Statistical downscaling derives relationship between the small scale variables and the large scale variable using statistical methods, e.g. analogue methods, regression analysis, and so on. Dynamical downscaling using regional climate model process the coarse resolution reanalysis data in more physical way. It is an appropriate way to simulate climate conditions in the future. Reanalysis data refers to the coarse resolution climate data that could be extended even a century into future: it is a combination of observation and model data through data assimilation procedure that is usually done by large climate centers. Furthermore, with the evolution of urban expansion and other land use change, studying their effect also requires the use of regional climate model.
Climate models. 1) Collaboration on the selection and aggregation of CMIP5 and CMIP6 models for an ensemble to represent future conditions
2) Collaboration on the downscaling of model results for flood, fire, and other peril applications
3) Collaboration on the scaling of future flood and fire conditions based on CMIP6 model ensembles, including estimations of uncertainties associated with the use of the CMIP6 model data
4) Collaboration on creation of bespoke scenario-driven new simulations with GISS models
