| description abstract | The ability to predict heavy rain and floods in Africa is urgently needed to reduce the socioeconomic costs of these events and increase resilience as climate changes. Numerical weather prediction in this region is challenging, and attention is being drawn to observationally based methods of providing short-term nowcasts (up to ∼6-h lead time). In this paper a freely available nowcasting package, pySTEPS, is used to assess the potential to provide nowcasts of satellite-derived convective rain rate for West Africa. By analyzing a large number of nowcasts, we demonstrate that a simple approach of “optical flow” can have useful skill at 2-h lead time on a 10-km scale and 4-h lead time at larger scales (200 km). A diurnal variation in nowcast skill is observed, with the worst-performing nowcasts being those that are initialized at 1500 UTC. Comparison with existing nowcasts is presented. Such nowcasts, if implemented operationally, would be expected to have significant benefits. | |
