Simulation of vegetation feedbacks on local and regional scale precipitation in West Africa

Abstract

Abstract Planned changes to land use in West Africa have been proposed to both combat desertification and to preserve biodiversity in the region, however, there is an urgent need for tools to assess the effects of these proposed changes on local and regional scale precipitation. We use a high-resolution, convection-permitting numerical weather prediction (NWP) model to study how the initiation and propagation of mesoscale convective systems (MCS) depends on the surface vegetation cover. The simulations covered a 4-day period during the West African monsoon in August 2006. In many aspects of the simulations, there was evidence of vegetation type exerting a significant influence on the location of precipitation where the influence of orography and coastal water was minimal. In this study, vegetation was classified according to the fractional coverage of tree (>30%) and grass (>30%) plant functional types. Tree-grass boundary cover was defined where more than 3 grid cells of both tree and grass occurred in a moving 3 × 3 window, which was further enlarged using a 3 grid cell (∼12 km) buffer. We found that over the whole study region (5N to 17N and 11W to 9E) 33.8% of convective initiations occur over tree-grass boundaries that cover only 28.4% of the land surface. This is significantly more than would be expected by chance (p = 0.0483), providing support to the hypothesis that vegetation gradients provide heat and moisture gradients, of a similar magnitude to that of soil moisture. Additionally, we found that on average, more time under an MCS occurred over boundary cover and orography, followed by tree cover, during the afternoon and evening period, thus supporting the hypothesis that land cover type influences the location of larger propagating systems. Contrasting patterns were found in the quantity of precipitation between small-scale convective cells and larger scale MCS. More small-scale precipitation accumulated, on average, over grass cover during the afternoon period, indicating a tendency for small-scale convection, initiated over boundaries, to prefer the drier and warmer grass side of vegetation boundaries in the afternoon period. However, once these smaller scale convective cells merge together to form larger MCS, a tendency for the most intense precipitation to fall over tree cover was observed. When intense precipitation (>10 mm per hour) occurred simultaneously over tree, boundary and grass cover, we found the highest precipitation rate to be most frequently over tree cover (48.4%), and least frequently over boundary cover (19.9%), indicating a preference of MCS for cooler, more moist forest cover. These results show for the first time that convection-permitting NWP models do exhibit responses to vegetation similar to those observed in the real world, and therefore are useful tools to assess the impacts of proposed future land use changes.