Abstract—Hydrometeorological disasters such as flooding in urban areas are a big problem that must be managed. Indonesia as part of the maritime continent, has high rainfall variability, both temporally and spatially. Unfortunately, the density of instruments for measuring rainfall is still low. To solve the problem, this research will try to utilize and modify Closed Circuit Television (CCTV) cameras which have a large number in terms of quantity as instruments for measuring rainfall. The purpose of this research is to obtain rainfall image information and data generated by CCTV cameras. The image data is converted to quantitative rainfall data. The method used is the K-NN algorithm and machine learning. The research location is located in a corner of the city of Bandung with a geographical position of 60 53”30.49'S and 107.035” 12.27' E. The results of this research show that the K-NN algorithm can be applied to estimate rainfall data from CCTV images with an accuracy of more than 98%. The level of accuracy generated between CCTV camera image data and AWS is 94%. The level of accuracy is high means that CCTV camera image data can represent or be converted into quantitative rainfall data. Index Terms—Rainfall, Rain Gauge, CCTV Camera, Image Processing, Validation.
Copyrights © 2023