Sadewa satellite remote sensing data to Manggarai 1-hour water level machine learning model

Abstract

The Manggarai Water Gate is a measurement point strategically located to measure Jakarta's flooding magnitude that keeps increasing from year to year. The 2015s gate capacity improvement underscores this importance. This paper applies a machine learning model that utilizes an atmospheric approach to predict the Manggarai water level as output. In the process, optimization is done by comparing three spatial input sizes and performing a sensitivity analysis of the input variables. Using a simple recurrent sequence, the model can predict the water level with a coefficient of determination (${R}^{2}$) reaching 0.7 using 18-hour recurrent data. This study can be used as the basis for further development that can take satellite data lead time advantage that is crucial for the early warning system.