Land-atmosphere Interactions in Climate Models for Medium-range Applications in Eastern Africa

Abstract

Climate models are valuable tools that aid in understanding the evolution of local to global climate systems and possible impacts on society across multiple timescales. Over Eastern Africa, climate variability is double-faced, presenting both hazards and opportunities. As the human population density increases in high-potential areas, the level of vulnerability of communities to climate hazards rises. The increase in human population, together with government efforts to establish new infrastructure such as roads and mega dams, contributes to significant modification of the natural land cover. Such changes are known to introduce local feedbacks to the climate system that could be responsible for regional to remote influence on the prevailing weather. Although advances in climate modeling have presented opportunities for research in simulating the local scale processes, the computing resources required for explicit representation of such processes in dynamical models are lacking in Eastern Africa. In this regard, most research and operational centers rely on parameterization of key weather and climate processes. Of importance, the human footprint on the climate system, especially in under developed and developing countries can be traced to land use changes. With the introduction of feedbacks in the local scale, it is not yet quantified how such processes modify the climate from local to regional scales. In this study, attention is given to understanding land-atmosphere coupling processes over Eastern Africa. Eight global climate models (GCMs) participating in the Coupled Model Intercomparison Project Phase 6 (CMIP6) were selected based on similar physics, initialization and forcings. The models were used for 36-year historical period, 1979-2014, to provide benchmark results on land-atmosphere coupling, based on the terrestrial and atmospheric pathway metrices. Further, identification of dry, wet or transitional soil moisture regimes was based on locally weighted scatterplot smoothing technique. The trend in land use changes, which is a key factor in triggering feedbacks to the climate system, was studied using the LUH2 dataset. In addition, widely used rainfall datasets, comprising of CHIRPS, TAMSAT, ARC2 and PERSIANN, together with ERA5, were used to identify extreme events for the month of April, based on box-and-whisker analysis. This formed the basis for case study simulation using the Weather Research and Forecasting (WRF) model. WRF simulations focused on investigating the variability of rainfall and temperature with topography, feedbacks due to land use changes, and detection of soil moisture-induced convection.