On the Development of the Impact-Based Forecast Model in Indonesia

Considering the frequency of hydrometeorological disasters in Indonesia, it has been statistically proven to be the most common type of disaster. Consequently, the Indonesian Agency for Meteorology Climatology and Geophysics (BMKG) is compelled to shift its paradigm of weather forecasting toward from the conventional forecast to the impact-based forecast (IBF) approach. While BMKG's weather information is generally accurate, a critical issue arises: the identification of high-risk areas for disasters proves to be challenging using conventional information. To derive the risk index, we calculate potential hazard, capacity, and vulnerability data from InaRISK, a resource provided by the Indonesian National Disaster Management Agency (BNPB). Subsequently, we calculate the probability of various impact values occurring using the ensemble IFS 0.125 model with a time-lagged methodology. The final assessment is attained through a cross-analysis technique between impact and likelihood, resulting in a range of values from − 6 to 10 (excluding 0). Values from 1 to 10 serve as primary information in the IBF matrix, classifying warnings into yellow, amber, and red. The model’s performance was verified using a dichotomous method, comparing its predictions to 172 recorded disastrous events. The IBF successfully predicted 74% of these events, particularly excelling in the severe and significant event categories. Nevertheless, it also underscores a significant issue with categorization accuracy within the minor and minimal categories, as the majority of predicted events did not align with their assigned categories.