Multi-Temporal Performance Evaluation of Satellite Precipitation Products over Complex Orographic Terrain

Abstract

Ground-based rain gauge networks in complex orographic regions such as Bogor, Indonesia, often suffer from sparse distribution and spatial gaps, limiting effective weather monitoring and disaster mitigation. This study evaluates the accuracy of four satellite precipitation products—gpm-imerg final run, chirps, gsmap_nrt, and persiann-ccs—against daily observations from 12 tipping bucket and AWS stations during December 2023–November 2025. Data processing employed Python, Google Earth Engine, and Quantum Geographic Information System with UTC-to-WIB synchronization, nearest-neighbor point-to-grid extraction, and multi-temporal aggregation at daily, 10-daily, monthly, and seasonal scales. Performance was assessed using correlation coefficient R, Root Mean Square Error, Mean Absolute Error, regression analysis, and categorical metrics including Probability of Detection, False Alarm Ratio, Critical Success Index, and Frequency of Hits. Results show weak daily correlations (R < 0.20) due to local convective dynamics and orographic lifting, but accuracy improves at aggregated scales (10-daily R = 0.536–0.718; monthly R = 0.685–0.777). gpm-imerg consistently outperformed others, achieving the highest monthly correlation (R = 0.777), lowest daily error (Root Mean Square Error = 14.92 mm/day), and optimal detection capacity. persiann-ccs tended to overestimate rainfall in high terrain, while gsmap underestimated systematically. Findings suggest uncorrected daily satellite products should be applied cautiously, with gpm-imerg recommended for gap filling and water resource assessments at 10-daily resolution or bias-corrected for daily operations.