Multi-Model Assessment of Regional Climate Simulations over Pakistan Using SMHI-RCA4 Downscaled GCMs under CORDEX framework

Abstract We present an evaluation of ten Coordinated Regional Climate Downscaling Experiment (CORDEX) RCA4 global climate models (GCMs) in producing the comparable precipitation and temperature simulations with Climate Research Unit (CRU) datasets (from: 1975–2005) for generating future climate ensemble scenarios over Pakistan. The driest precipitation regime is depicted in the south west of Pakistan for which GFDL performed better. CanESM2, EC-Earth, CSIRO and GFDL produce realistic temperatures when compared to CRU. All models underestimate temperatures compared to CRU. EC-Earth, MPI and NorESM1 show <10% bias in precipitation over the south east of Pakistan. Model correlations for precipitation are highest in eastern parts of Pakistan (> 0.5). Over south west of Pakistan NorESM1 performs better than the rest of the models. In the HKH region, the biases are smaller in the western Hindukush-Karakoram-Himalaya (HKH) and larger wet biases (up to 100%) in the eastern part (Himalayas). Below 35o N, for mean temperature, CSIRO performs better. Over the HKH region EC-Earth shows relatively lower biases. Climatological trends are captured by EC-Earth only, however it underestimates precipitation over the HKH. Model ensemble shows a change of 6o C under RCP 8.5 over Pakistan while RCP 4.5 stabilizes at 3o C by the end of the century. The future scenarios depict that HKH is experiencing higher variability is temperature with a decreasing trend in precipitation, while below 35o N a higher variability in precipitation and lower in temperature is seen.