Daily precipitation analysis using the CLARIS-LPB models over hydrological basins and climate regions within the CORDEX South America domain
Multiyear simulations of the South America daily precipitation are derived by six regional models and a stretched-grid general circulation model within the framework of the EU FP7 CLARIS-LPB Project. The different institution and models which participated in the coordinated project are SMHI RCA, MPI-M REMO, UCLM PROMES, USP RegCM3, CIMA MM5, IPSL LMDZ and INPE ETA. The setup of the different models attempts to follow the WCRP CORDEX protocol closely such as the horizontal resolution of about 50 km and size of the domain, which covers the whole continent of South America. These models have been integrated for the period of 1989-2008 using the ECMWF ERA-Interim reanalysis as initial and lateral boundary conditions. The daily precipitation observational dataset used is from the NOAA Climate Prediction Center (CPC) with a resolution of 0.5 degree. The precipitation estimates from the high resolution TRMM are also used to ascertain observational uncertainties. The skill of the models and ensemble are evaluated using the probabilistic distribution function over twelve regions with different hydrological and climate characteristics. The seven regions representing hydrological basins are South Amazon, Northeast Brazil, South Atlantic Convergence Zone, Paraguay, Upper Parana, Lower Parana and Uruguay. The other five regions have tropical humid, tropical wet-dry, dry semi-arid, subtropical humid and temperate oceanic climate based on the Koeppen-Trewartha classification. The regions are established to evaluate the simulated daily precipitation at areas of similar characteristics. The seasonal means and extreme indices based on percentile approach are also calculated The models have a relatively high skill in simulating the wintertime precipitation at the different regions. During summer, when the precipitation is mainly driven by convective activity, most of the models have relatively low skill. Attempts are made to explain the different biases among each region due to the different convective parameterization used. The convective schemes used in the models are Kain-Fritsch, Tiedtke with Nordeng enclosure, Grell, Emanuel and Betts-Miller. The aim of this study is to characterize the uncertainty in simulating the South America climate in terms of daily precipitation distribution, mean and extreme values.
AGU Spring Meeting Abstracts
- Pub Date:
- May 2013
- 0300 ATMOSPHERIC COMPOSITION AND STRUCTURE