Google Earth Engine–Based Landslide Early Warning System (GLEWS) for Rainfall-Induced Landslides in Idukki, India

We have developed a Google Earth Engine–based landslide early warning system, named GLEWS, designed to issue landslide alerts for Idukki. The GLEWS uses 3-hour global precipitation measurement (GPM) integrated multi-satellite retrievals for GPM precipitation data and hourly global forecast system rainfall forecasts in combination with a slope stability factor of safety (FoS) map at 30 m resolution, where FoS ≤ 1.5 indicates high susceptibility and FoS > 2.0 indicates stability. The system integrates live rainfall telemetry from eight automatic weather stations to calibrate satellite precipitation using conditional merging, combining an inverse distance weighting–interpolated error field with gauge observations. A critical daily rainfall of 213 mm and a critical antecedent rainfall of 2,921 mm separate extreme from non-extreme precipitation. Alerts are issued as red (extremely critical), orange (very critical), yellow (critical), and green (not critical) based on rainfall exceedance and slope FoS. In the August 15, 2018, case study, GLEWS achieved 72.6 percent sensitivity with conditional merging using five India Meteorological Department (IMD) point-gauge observations with partial gauge correction, compared with 8.9 percent using uncalibrated or IMD 0.25° gridded precipitation, demonstrating that point-level gauge observations are essential for operational performance. A second case study using the July 7, 2023, Munnar Gap Road landslide showed that conditional merging increased the daily rainfall estimate from 20.0 mm (raw GPM) to 63.9 mm, a threefold improvement. The system has been ported from JavaScript to Python enabling automated gauge ingestion and autonomous operation during monsoon seasons.