Evaluation of TAMSAT satellite rainfall estimates for southern Africa: A comparative approach

Abstract

Developing countries with a paucity of direct rainfall measuring systems on the ground, satellite remote sensing is increasingly being relied upon as a means through which national governmental response and future strategies to manage natural disasters, such as drought and flooding are decided upon. The aim of this study was to assess the performance of the Tropical Applications of Meteorology Using Satellite Data and Ground-Based Observations (TAMSAT) monthly rainfall dataset for southern Africa by comparing it to two widely used stable global rainfall datasets, the Global Precipitation Climatology Centre (GPCC) and the Global Precipitation Climatology Project (GPCP) datasets. Using rainfall records for the period 1984–2010, four statistical measures namely, the correlation coefficient, the Root Mean Square Error (RMSE), relative bias, mean absolute deviation and the Nash Sutcliffe Efficiency index were calculated to facilitate pair-wise comparisons. This inter-product comparison study found a good agreement between TAMSAT satellite rainfall and the two global rainfall products with regard to the spatial and temporal representation of monthly rainfall over the region. However, it was found that TAMSAT consistently provides lower monthly rainfall estimates compared to both GPCC and GPCP rainfall estimates over southern Africa. The relative bias between TAMSAT and the two global rainfall products (GPCC and GPCP) was much lower (≤10%) during the austral summer months (November–March) compared to winter when the relative bias surpassed 70%. These findings indicate TAMSAT is geared more towards convective precipitation brought about by the inter-tropical convergence zone in summer. The observed large discrepancy between TAMSAT estimates and those from two mature global rainfall products during winter raises a key concern regarding the use of TAMSAT as a tool for monitoring rainfall patterns and volumes across the southern reaches of the African continent. Overall, the findings of this study demonstrate the diminished ability for TAMSAT to “see” non-convective precipitation from frontal systems as well as orographic rainfall due to its reliance upon the observation of the cold cloud duration and convection characteristics of tropical rainfall systems. This calls for the need to further optimise the TAMSAT algorithm using gauge data so that TAMSAT is better able to characterise different precipitation events in those regions where convective rainfall is far from the only major source of rainfall.