Evaluation of the weather, Research and Forecasting Model over Tunisia for the Study of Drought

Located in the subtropical semi-arid climate and being prone to water scarcity and irregularity, Tunisia is a country in North Africa particularly vulnerable to the potential impacts of rapid climate change. The interannual climate variability already poses severe impacts on rain-fed agriculture through recurrent and extensive drought episodes. These impacts mainly target the rural societies, approximately 50 percent of the total population, who are dependent on the cultivated and arable land in the country. Drought monitoring and early warning systems are the first steps to approach the drought management and adaptation strategies that can alleviate the impacts, if delivered to decision makers in appropriate temporal and spatial scales. Current climate projections have been built on global climate models outputs with a spatial resolution of 100-200 km, which is not sufficient for regional impact studies (typically 4-12 km) considering the heterogeneous topography and climatic gradients in the region. Dynamical downscaling through regional climate models is an acknowledged approach to obtain high resolution data appropriate for impact studies. Knowing that each specific part of the world has its own unique climate characteristics, the selection of the best model parameter configuration with which to generate realistic outputs is a major challenge with regional climate models. Using the climate version of the Weather, Research and Forecasting (WRF) model, ERA-Interim reanalysis data and a high resolution SST dataset (5 km), we evaluated the performance of 7 different configuration physics over a domain that included Tunisia over a one-year period. These one-year simulations were compared to the high-resolution gridded dataset of CHIRPS and station measurements for precipitation. Our results show that cumulus parameterization scheme has the most pronounced impact on the precipitation and especially in the northern part of the country where is under the influence of Mediterranean climate. Implementing the selected configuration, based on the results of this research, enables us to reduce the WRF biases in longer-term climate change simulations.