Providing an accurate and a timely daily weather forecast on regular basis is necessary but require a very expensiveinfrastructure especially when it relies on numerical weather prediction (NWP) technology. On the other hand, the needsto deliver probabilistic forecast product will require many ensemble-runs of the NWP model, leading to more computingtime or power that will considerably increase computing costs. This study aims to test the performance of alternativeapproach for a regularly-scheduled daily weather forecast in Indonesia involving many ensemble members. It focuses onevaluating the forecast of daily rainfall occurrence operated under a relatively low computing cost by combining dynamicalapproach and bias correction technique. The dynamical approach is conducted by performing Weather Research andForecasting (WRF) model simulation by using only one member of Global Forecast System (GFS) data as initial conditionand boundary condition data. Meanwhile, the bias correction technique is performed as alternative to get as many aspossible ensemble members that can be used for calculating probabilistic rainfall forecast. For this purpose, this study uses20 ensemble members of forecasted data obtained from Global Ensemble Forecast System (GEFS). The combineddynamical and bias correction technique has considerably reduced forecast data production time, so it is possible to makeit regularly-scheduled for daily updates. Evaluation on the forecast results show that the model forecasts are consistentwith GPM satellite data and obtained adequate skills (Brier Skill Score ~0.7), especially over land areas with low tomedium topography.
展开▼