The evaluation of weekly extended range river basin rainfall forecasts and a new bias correction mechanism for flood management in India

Operational extended range forecasts are being disseminated once every week by the India Meteorological Department (IMD) for several sectorial applications. These forecasts show a reduction in amplitude and variance as a function of lead-time. Such reductions in variance can be due to several physical factors: inherent forecast model bias, a problem relating to initial conditions, leaddependent statistical biases, etc. A week-by-week analysis shows that such biases are not systematic. Rainfall forecasts are underestimated in some regions, while others overestimate rainfall amplitude. To correct the bias in the extended range weekly averaged forecast, a statistical post-processing method (normal ratio correction) is proposed to make the outlook more valuable at a longer lead-time. The correction method is based on the World Meteorological Organization (WMO) technical guidance on rainfall estimation and is also shown to be useful for rainfall forecasts. In this analysis, we evaluate the extended range forecast skill at the river sub-basin-scale and show that there are several river sub-basins over the central Indian region where the correction has improved the model forecast in the one to two-week range. Although this analysis was tailored toward making the river basins and sub-basins of India more readily realizable for flood forecasters, it can be used for any administrative boundaries such as block, district, or state-level requirements.

---

Opracowanie rekordu ze środków MEiN, umowa nr SONP/SP/546092/2022 w ramach programu "Społeczna odpowiedzialność nauki" - moduł: Popularyzacja nauki i promocja sportu (2022-2023).