EVALUATION OF RAPIDLY DEVELOPING CUMULUS AREA INDEX IN DETECTING HEAVY RAIN IN BANDUNG REGENCY

• Main Author: Azura, Aulia
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/54841
• Institution: Institut Teknologi Bandung

Bandung Regency is an area with high convective activity that can even be repeated. This can cause flooding which results in both material and life losses. Rain observation and Weather Radar can detect the signal of cloud growth that can cause heavy rain. However, it turns out that the evacuation process need to take faster time. In previous study, there is a method to detect rain faster than surface rain observations, that is Rapidly Developing Cumulus Area Index. This study evaluate the use of RDCA (Rapidly Developing Cumulus Area) index in detecting heavy rain in Bandung Regency. By adapting the methods in previous studies, this study uses a logistic regression model to generate an RDCA index value associated with the potential of rain by utilizing Himawari-8 data. In addition, the evaluation of the model with the analysis of the confusion matrix and brier score was also carried out. The results showed that by doing these evaluations, model with probability threshold p>0 and p>0,2 had good results, while for the threshold p>0,4 had poor results. From these results it can be said that there is no clear relationship between the magnitude of the RDCA and the predicted rain, as well as in terms of distinguishing light-moderate rain and heavy rain. It also shows certain limitations in the use of the original RDCA index in the Bandung Regency area. Then, by modifying the original RDCA algorithm using a new band, IR1-IR2, the RDCA model becomes better in predicting rain especially in distinguishing between heavy rain and light-moderate rain