| Sumario: | AGRHYMET operates a regional Numerical Weather Prediction (NWP) system using the Weather Research and Forecasting (WRF) model on in-house High-Performance Computing (HPC) system. It downscales the Global Forecast System (GFS) outputs: from 0.25° resolution to 0.12° over West Africa and the Sahel. In this Info Note, the forecast quality of both the WRF and GFS models is evaluated and compared. Results show complementary strengths between the AGRYMET implemented-WRF and GFS models. The frequency bias in the forecast of heavy daily rainfall frequency from WRF is lower (26.17%) compared to the bias from GFS (73.83%), implying that underestimation or overestimation of heavy daily rainfall
occurrence of WRF is lower than GFS. For light to moderate daily rainfall (1 – 10 mm), GFS performs slightly better than WRF. Overall, GFS overestimates the frequency of daily rainfall events more than WRF. According to the performance in forecasting heavy rainfall (>25 mm), WRF is more suitable in developing flood forecasting services for West Africa and Sahel. Both WRF and GFS forecasts generally capture the main rainfall patterns. In particular, the WRF model performs well along the Guinean coast. Beyond forecast performance and accuracy, regional modelling initiated by AGRHYMET RCC-WAS contributed to strengthen the West Africa and Sahel capacity and strategic autonomy with inhouse and tailored NWP forecast products.
|
|---|