Speaker
Description
Accurate rainfall onset prediction is essential for agricultural resilience across Zanzibar, Dar es Salaam, and Iringa. This study evaluates the traditional 4-Season temperate model against RegCM4 (regional), CAN-SIAM (climatological), and ECMWF/IFS (global). Results reveal a significant predictive gap, with observed onset exhibiting strong interannual and regional variability, while traditional models—especially 4-Season and CAN-SIAM consistently show delayed onset biases of 6–13 days. Analysis of 2015–2025 rainfall patterns indicate increasing variability in the Masika season, driven by dynamic convective processes. ECMWF/IFS demonstrates the highest agreement with observations (R² = 0.525), though all models underestimate rainfall intensity. The 4-Season model is found unsuitable for tropical applications due to its inability to capture rapid monsoon onset dynamics. These findings highlight the need for dynamic, data-driven forecasting approaches to improve agricultural planning and early warning systems in Tanzania.
| Title | Evaluation of Traditional and Numerical Meteorological Models for Rainfall Onset |
|---|