Evaluation of the performance of the WRF 1-Dimensional Lake model over the East Africa Great Lakes

Abstract:

This study seeks to investigate the performance of the 1-Dimensional lake model coupled to WRF over East Africa. The Africa Great lakes exert a great influence on the climate of the region and a number of studies have shown how the lake influences the circulation and the total precipitation over the region. The lakes have highly variable depths, with Lake Victoria having an average depth of 40m and Lake Tanganyika a depth of 450m. The Lake model for WRF was tested and calibrated for the Great lakes, however it was not tested for tropical lakes. We hypothesize that the inclusion of a 1-dimensional lake will reduce the precipitation bias as compared to the WRF model without the lake model. In addition initializing the lake temperature using a vertical temperature profile that closes resembles the one over these lakes will greatly reduce the spin up time. The simulations utilized three nested domains at 36, 12 and 4km. The 4km domain is centered over Lake Victoria Basin, while the 12 km domain includes all the lakes in East Africa. The Tropical Rainfall Measuring Mission (TRMM) datasets are used in evaluating the precipitation, and the following statistics were calculated: root mean square error, standard deviation of the model and observations and mean bias.

The results show that the use of the 1-dimensional lake model improves the precipitation over the region considerably compared to an uncoupled model. The asymmetrical rainfall pattern is evident in the simulations. However using the default vertical temperature profile with a three-month spin up is not adequate to transfer heat to the bottom of the lake. Hence the temperatures are still very cold at the bottom. A nine-month spin up improves the lake surface temperatures and lake temperatures at the bottom. A two year spin up greatly improves the lake surface temperatures and hence the total precipitation over the lake. Thus longer spin up time allows for adequate heat transfer in the lake. Initializing the vertical temperature with a temperature that closes resembles the observed profile, reduces the spinup time and precipitation bias.