| Abstract Submission No. | ABS-4.1-0116 |
| Title of Abstract | Integrating Machine Learning and Google Earth Engine for Flood Management in the Lower Manair River Basin, Telangana |
| Authors | Guddeti Sravya Sai*, Rama Mohan Kurakalva, Chinmoy Rajkonwar |
| Organisation | CSIR-North East Institute of Science and Technology (NEIST) |
| Address | Geo-Science and Technology Division (GSTD), CSIR-North East Institute of Science and Technology, Jorhat, Assam, India Pincode: 785006 E-mail: sravyasai1994@gmail.com |
| Country | India |
| Presentation | Poster |
| Abstract | Floods are recurring natural hazards in the lower Manair River Basin, Telangana, posing significant threats to agriculture, infrastructure and local livelihoods. This study presents an integrated flood monitoring and management framework using Google Earth Engine (GEE), a cloud-based geospatial analysis platform. Multi-temporal datasets from Sentinel-1 SAR (10 m), Sentinel-2 MSI (10¿30 m) and MODIS (250 m) were utilized to detect, map and analyze flood dynamics across spatial and temporal scales. Flood inundation was delineated using SAR-based thresholding, while the Normalized Difference Water Index (NDWI), derived from optical imagery enhanced the accuracy of water body classification. A decadal assessment (2012-2022), combined with rainfall records and reservoir release data, identified recurring flood hotspots in low-lying agricultural regions of Karimnagar district. Peak inundation depths reached 1.5-2 m during intense monsoon events, notably in September 2019, when late-monsoon depressions lead to extreme rainfall and widespread flooding. Annual flood-affected areas ranged from 45 to 120 km², with downstream flooding strongly linked to rainfall events ≥120 mm/day and high-volume reservoir discharges. Hydrological modeling within the GEE environment enabled near real-time flood mapping, significantly reducing processing time from several hours in traditional desktop systems to under 20 minutes. Accuracy assessment indicated reliable performance, with mapping range from 88 to92%. The study demonstrates potential of GEE-driven flood assessment as rapid, scalable and cost-effective tool to support disaster risk reduction, early warning systems and timely interventions, ultimately contributing to enhancing resilience in flood-prone communities of Telangana. |
| Keywords | Google Earth Engine, Machine Learning, Lower Manair River Basin, flood assessment, SAR, NDWI, Sentinel-2 |