Rainfall threshold analysis using satellite data for landslide events in Karanganyar Regency in 2024

Abstract Karanganyar Regency is a hilly region prone to landslides, especially during the rainy season. High rainfall, both in terms of daily intensity and multi-day accumulation, is one of the main triggering factors for landslides. This study aims to analyze the rainfall threshold that has the potential to trigger landslide events using CHIRPS satellite imagery as the primary source for rainfall estimation. The methodology includes statistical analysis using Pearson correlation between monthly rainfall and landslide occurrences, as well as an evaluation of the effectiveness of rainfall accumulation over 3, 5, 10, and 15-day periods in detecting landslide events. Evaluation metrics used include sensitivity, accuracy, and positive predictive power. The results indicate a very strong and statistically significant correlation between rainfall and landslide occurrences (r = 0.907; p < 0.001). While the 10-day rainfall accumulation period shows relatively better performance than other periods, the overall model performance remains low, with an overall accuracy value below 20%. These findings suggest that although CHIRPS data can provide an early indication of potential landslide events, further development is needed to improve the reliability of rainfall-based early warning systems in high-risk areas such as Karanganyar.