| Summary: | Future projections of the Greenland ice sheet melt are based on General Circulation Model (GCM) simulations. In particular, the reliability of downscaling methods forced by these simulations depends on the quality of the atmospheric circulation simulated by GCMs. Therefore, it is essential to analyse and evaluate the GCMs modelled general circulation for current climate (1961-1990). Atmospheric circulation type classifications offer a very interesting approach for evaluating the GCM-based circulation at a daily time scale compared to the most used methods based only on monthly means. Indeed, the circulation type classification allows a precise and detailed analysis of each circulation type and so, it gives much more information on the ability of GCMs to simulate the different circulation types and consequently the climatic variability of a region. In fact, exceptional circulation events over Greenland, which cannot be taken into account by the monthly mean approach, have much more impact on the melt than the mean atmospheric state. Thus, an automated correlation-based atmospheric circulation type classification (CTC) is used for evaluating the new GCM outputs (available on http://cmip-pcmdi.llnl.gov/cmip5/) computed for the upcoming IPCC report (AR5). The daily geopotential height at 500 hPa simulations of the GCMs for current climate are compared to the NCEP-NCAR 1 and the ECMWF reanalysis data for the summer months (JJA), when melt is the most important. To achieve this, the classification is first done for the reanalysis data over 1961-1990 and afterwards, the types of the reanalysis based CTC are imposed for classifying the GCM datasets over 1961-1990 (from the historical experiment) to allow a direct type per type comparison based on the frequency distribution of each dataset. This approach also gives the opportunity to study the intraclass repartition differences between the reanalysis and the GCMs. After the evaluation of the GCM simulations for current climate, the future projections driven by RCP ...
|
|---|