|
Rapid Monitoring of Flood Events in Geothermal Surrounding Areas Using Machine Learning and Multi-sensor Data |
|---|---|
| รหัสดีโอไอ | |
| Creator | 1. Fahmi Arif Kurnianto 2. Elan Artono Nurdin 3. Era Iswara Pangastuti 4. Moch Ilham Deka Rafsanjani |
| Title | Rapid Monitoring of Flood Events in Geothermal Surrounding Areas Using Machine Learning and Multi-sensor Data |
| Publisher | Thai Society of Higher Education Institutes on Environment |
| Publication Year | 2567 |
| Journal Title | EnvironmentAsia |
| Journal Vol. | 17 |
| Journal No. | 2 |
| Page no. | 78-90 |
| Keyword | Rapid monitoring, Flood, Machine learning, Multi Sensor Data |
| URL Website | http://www.tshe.org/ea/index.html |
| Website title | EnvironmentAsia |
| ISSN | 1906-1714 |
| Abstract | Volcano arcs, particularly the Sunda Arc, are predominant in Indonesia and constitute a key tectonic system in current geodynamics, making it essential for several stakeholders to consider geothermal exploration while taking into account the environmental impact. In previous reports, conventional methods were used in mapping environmental impact zones, leading to inefficiency in the monitoring process. Therefore, this study aimed to monitor flood in geothermal surrounding areas using machine learning and multisensor data. In the process, sentinel 1 GRD was adopted to map flood area distribution and was compared to Sentinel 2 optic data. Furthermore, Google Earth Engine was used to process several steps, including acquisition, polarization, measuring indices, and post-processing. Mapping was conducted at flooded areas in Ciwidey and Pasirjambu as geothermal surroundings. The results showed that flooded areas were located in the built-up areas as a part of watershed downstream. Ciwidey District had a wider flooded area than Pasirjambu. In June 2022, VH polarization of flooded areas in Ciwidey was recorded at -25, correlating with field data. Meanwhile, the comparison between VH polarization in Pasirjambu and Sentinel 2 indicated that flooded areas were associated with low NDVI and SAVI values, representing low infiltration. Google Earth Engine was capable of detecting several areas with different topography through optical and SAR data fusion. Differences in flood detection were discovered between VH and VV polarizations in the subset area around geothermal area. VH polarization was more accurate for detecting flood distribution due to the capability to acquire inundation data in densely vegetated areas. Additionally, the method can be used to distinguish permanent water bodies in the form of lakes/craters from flood inundations. This study concluded that machine learning and multi-sensor data facilitated the mapping and monitoring of flood distribution using both temporal analysis and a faster process. |
