Evaluation of the UK Met Office's HadGEM3-RA and HadRM3P regional climate models within South America-CORDEX simulations: ENSO related interannual precipitation variability

Thursday, 18 December 2014
Maisa Rojas, University of Chile, Santiago, Chile and Deniz Bozkurt, University of Chile, Center for Climate and Resilience Research, Santiago, Chile
This study aims to investigate and compare the ability of the UK Met Office's HadGEM3-RA and HadRM3P regional climate models (RCMs) to simulate mean and interannual variability of precipitation over South America with a special focus on Chile. The HadGEM3-RA is a regional version of the newly developed HadGEM3 global model and the HadRM3P is based on the earlier HadCM3 global model. The RCMs simulations were carried out at 0.44o x 0.44o degree resolution over South America-CORDEX domain for the period 1989-2008. The initial and boundary conditions were provided by ERA-Interim Reanalysis data available at 6-h intervals with a resolution of 1.5o x 1.5o in the horizontal and 37 pressure levels. We compare the results against a number of observational datasets, including gridded dataset of CRU, UDEL, TRMM and GPCP. Moreover, available station data is derived from Direccion General de Aguas (DGA) mainly for Central Chile, which is the heartland of Chile with the highest population and important economic activities. The analysis is mainly focused on evaluating the abilities of the RCMs in simulating spatial pattern and ENSO related precipitation variability in different subregions of South America-CORDEX domain. In general, both RCMs have a good skill in reproducing spatial pattern and annual cycle of observed precipitation in climatically different subregions. However, both RCMs tend to underestimate precipitation in the Amazon Basin, which is more pronounced in the HadRM3P simulations. On the contrary, the RCMs tend to overestimate the precipitation over the Andes and southern Chile. The overestimation could be related to the physical core of the RCMs, but the discrepancies could also arise due to insufficient station network, especially in the mountainous areas, potentially yielding smaller precipitation quantities in the observed data than the true ones. In terms of interannual variability, the models capture ENSO related wet and dry interannual precipitation anomalies with some differences in the magnitude of precipitation. The study will be also extended with evaluation the ability of the RCMs and driving fields to simulate large-scale atmospheric fields in order to understand the biases in the regional simulations. Moreover, intercomparison of other South America-CORDEX simulations will be also included.