Delivering local-scale climate scenarios for impact assessments
Seminar Room 1, Newton Institute
Process-based models, used in assessment of impact of climate change, require daily weather as one of their main inputs. The direct use of climate predictions from global or regional climate models could be problematic, because the coarse spatial resolution and large uncertainty in their output at a daily scale, particularly for precipitation. Output from a climate model requires application of various downscaling techniques, such as weather generator (WG). WG is a model which, after calibration of site parameters with observed weather, is capable of simulating synthetic daily weather that are statistically similar to observed. By altering the site parameters using changes in climate predicted from climate models, it is possible to generate daily weather for the future. A dataset, ELPIS, of local-scale daily climate scenarios for Europe has been developed. This dataset is based on 25 km grids of interpolated daily precipitation, minimum and maximum temperatures and radiation from the European Crop Growth Monitoring System (CGMS) meteorological dataset and climate predictions from the multi-model ensemble of 15 global climate models that were used in the IPCC 4th Assessment Report. The site parameters for the distributions of climatic variables have been estimated by the LARS-WG weather generator for nearly 12 000 grids in Europe for the period 1982Ė2008. The ability of LARS-WG to reproduce observed weather was assessed using statistical tests. This dataset was designed for use in conjunction with process-based impact models (e.g. crop simulation models) for the assessment of climate change impacts in Europe. A climate scenario generated by LARS-WG for a grid represents daily weather at a typical site from this grid that is used for agricultural production. This makes it different from the recently developed 25 km gridded dataset for Europe (E-OBS), which gives the best estimate of grid box averages to enable direct comparison with regional climate models.