IMAGE Danube basin rainfall description Capturing
IMAGE Danube basin rainfall description
Capturing the spatial and temporal correlation of multiple variables in a weather generator is challenging. IMAGE is a massively multi-site, multivariate daily stochastic weather generator. It models temperature and precipitation variables as latent Gaussian variables with temporal behaviour governed by an auto-regressive model whose residuals and parameters are correlated through resampling of principle component time series of empirical orthogonal function modes. IMAGE can be used on any weather variable and could also be applied to non-weather variables for which spatial correlation is important (eg river level, normalized vegetation difference index (NDVI)). IMAGE is capable of generating long time series (> 10,000 years) of synthetic weather data very quickly and using modest computational resources. The data produced by IMAGE can be used directly as the hazard module of a catastrophe model, or alternatively can be used as input to a second hazard model such as a hydrological model for river flooding. Methodology Each dataset is 10,000 years of synthetic data generated for 1492 grid cells within the Danube river basin. Driving data was generated by the as part of the EURO-CORDEX project. The data have been bias-corrected using E-OBS earth observations data (http://www.ecad.eu/download/ensembles/download.php). Data have been generated for five different time periods: 1970-1999; 2020-2049 and 2070-2099 under RCP 4.5 scenario; and 2020-2049 and 2070-2099 under RCP 8.5 scenario. For each time period four different bias-corrected CORDEX simulations are available. These are the MPI CSC REMO 2009 regional climate model driven by the MPI-M Earth System Model general circulation model, the RCA4_v1 regional climate model driven by the Met Office Hadley Centre Earth System Model general circulation model, the RCA4_v1 regional climate model driven by the EC-EARTH general circulation model and the RACMO regional climate model driven by the EC-EARTH general circulation model. Resolution The data is provided at 0.25° resolution on a regular latitude-longitude grid.
Data Information CSV Sample Data: Each row represents one day in the synthetic time series with a rainfall value in millimetres for each grid cell in the domain. A separate file contains the latitude and longitude of each grid cell. NetCDF Complete Dataset: The full 10,000 years of data is available on request through Oasis Hub in NetCDF format, split into two files of 5,000 years each to allow for manageable file sizes.
Contact : Department of Physics Dr. Stephen Hardwick E-mail : [email protected] Imperial College London SW7 2AZ, London United Kingdom
Capturing the spatial and temporal correlation of multiple variables in a weather generator is challenging. IMAGE is a massively multi-site, multivariate daily stochastic weather generator. It models temperature and precipitation variables as latent Gaussian variables with temporal behaviour governed by an auto-regressive model whose residuals and parameters are correlated through resampling of principle component time series of empirical orthogonal function modes. IMAGE can be used on any weather variable and could also be applied to non-weather variables for which spatial correlation is important (eg river level, normalized vegetation difference index (NDVI)). IMAGE is capable of generating long time series (> 10,000 years) of synthetic weather data very quickly and using modest computational resources. The data produced by IMAGE can be used directly as the hazard module of a catastrophe model, or alternatively can be used as input to a second hazard model such as a hydrological model for river flooding. Methodology Each dataset is 10,000 years of synthetic data generated for 1492 grid cells within the Danube river basin. Driving data was generated by the as part of the EURO-CORDEX project. The data have been bias-corrected using E-OBS earth observations data (http://www.ecad.eu/download/ensembles/download.php). Data have been generated for five different time periods: 1970-1999; 2020-2049 and 2070-2099 under RCP 4.5 scenario; and 2020-2049 and 2070-2099 under RCP 8.5 scenario. For each time period four different bias-corrected CORDEX simulations are available. These are the MPI CSC REMO 2009 regional climate model driven by the MPI-M Earth System Model general circulation model, the RCA4_v1 regional climate model driven by the Met Office Hadley Centre Earth System Model general circulation model, the RCA4_v1 regional climate model driven by the EC-EARTH general circulation model and the RACMO regional climate model driven by the EC-EARTH general circulation model. Resolution The data is provided at 0.25° resolution on a regular latitude-longitude grid.
Data Information CSV Sample Data: Each row represents one day in the synthetic time series with a rainfall value in millimetres for each grid cell in the domain. A separate file contains the latitude and longitude of each grid cell. NetCDF Complete Dataset: The full 10,000 years of data is available on request through Oasis Hub in NetCDF format, split into two files of 5,000 years each to allow for manageable file sizes.
Contact : Department of Physics Dr. Stephen Hardwick E-mail : [email protected] Imperial College London SW7 2AZ, London United Kingdom
