High-resolution dynamic downscaling to assess the impacts of climate change in regions of complex topography -

Abstract

High resolution dynamical downscalling is conducted for the past and the near future (untill 2050) climates under two Representative Concentration Pathways (RCP4.5 and RCP8.5) over a complex topographical terrain along the eastern Mediterranean, with a focus on Lebanon, using the Weather Research and Forecast (WRF) model forced by the High-Resolution Atmospheric Model (HiRAM). For this purpose, WRF performance for the study area was first evaluated by sequentially downscaling Global Forecast System (GFS) Reanalysis (resolution 1°) model results, for a mild and wet year and a hot and dry year, to three local horizontal resolutions of 9, 3 and 1 km. Simulated near-surface hydrometeorological variables were compared at different time scales against data from an observational network over the study area. The overall performance of WRF at 1 and 3 km horizontal resolution was very satisfactory, with significant improvement over the 9 km downscaling simulation. The findings therefore indicated that a 3km resolution is sufficient for the downscaling, especially that it would allow more years and scenarios to be investigated compared to the higher 1km resolution at the same computational effort. Then, a set of ten downscaling simulations at 3 km resolution were performed using WRF driven with HiRAM, to generate future climate projections of annual and seasonal temperature and precipitation changes over Lebanon. Two past years (2003 and 2008), were simulated to evaluate the model and to serve as a baseline scenario. The downscaled data were in the range of recent observed climatic variability, and therefore corrected for the cold bias of HiRAM. Eight future years were then selected based on an anomaly score that relies on the mean annual temperature and accumulated precipitation to identify the worst year per decade from a water resources perspective. One year per decade, from 2011 to 2050, and per RCP was simulated. The results indicated that by the middle of the century, the study area might be exposed to a sign