Montiel et al. (2026) Evaluation of satellite precipitation products across climatic and topographic gradients in a basin in Northern South America

Identification

• Journal: Environmental Challenges
• Year: 2026
• Date: 2026-02-07
• Authors: Jhonny I. Pérez Montiel, Adriana M. Choles Epiayu, Cristian Diaz Moscote
• DOI: 10.1016/j.envc.2026.101426

Research Groups

• Grupo de Investigación GISA, Facultad de Ingeniería, Universidad de La Guajira, Riohacha, Colombia

Short Summary

This study evaluates the performance of five gridded precipitation satellite products (GPPs) across climatic and topographic gradients in the Ranchería river basin, northern Colombia. The findings indicate that CHIRPSv3 exhibits the best overall performance and inter-scale consistency, with all products showing improved reliability with temporal aggregation but degraded performance at higher elevations.

Objective

• To evaluate the performance of five gridded precipitation satellite products (CHIRPSv2, CHIRPSv3, ERA5-Land, MSWEPv2.8, and PERSIANN–CDR) in the Ranchería river basin, northern Colombia, as a function of temporal scale, topographic gradients, and precipitation intensity.

Study Configuration

• Spatial Scale: Ranchería river basin, northern Colombia (approximately 4267 km²), ranging from 0 m (sea level) to 3875 m above sea level in the Sierra Nevada de Santa Marta. Evaluation was conducted using 14 rain gauges across topographic gradient intervals: [0, 50] m, (50, 100] m, (100, 150] m, (150, 200] m, (200, 250] m, (250, 450] m, and (450, 900] m above sea level.
• Temporal Scale: Evaluation period from 1991 to 2020, analyzed at daily, 3-day, monthly, and annual time scales.

Methodology and Data

• Models used: CHIRPSv2, CHIRPSv3, ERA5-Land, MSWEPv2.8, PERSIANN–CDR.
  • Spatial resolutions: CHIRPSv2/v3 (0.05°), ERA5-Land/MSWEPv2.8 (0.10°), PERSIANN–CDR (0.25°). All GPPs were resampled to 0.05° spatial resolution.
• Data sources:
  • Reference data: 14 rainfall stations from the Colombian Institute of Hydrology, Meteorology, and Environmental Studies (IDEAM) national monitoring network.
  • GPPs:
    • CHIRPSv2/v3: Hybrid products integrating infrared (IR) estimates, high-resolution climatology (CHPclim), and in situ stations. CHIRPSv3 uses significantly more rain gauge records than v2.
    • ERA5-Land: Reanalysis product from ECMWF, combining meteorological observations with physical models.
    • MSWEPv2.8: Merged product incorporating satellite, reanalysis, and rain gauge observations (e.g., GHCN, GSOD, LACA&D).
    • PERSIANN–CDR: Satellite-based estimates using artificial neural networks (ANNs) on remotely sensed data, with monthly GPCP rain gauge data for climate consistency.
  • Metrics: Continuous (Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Pearson correlation coefficient (r), Nash-Sutcliffe efficiency (NSE), Percentage Bias (PBIAS), Index of Agreement (DR), Kling-Gupta efficiency (KGE)) and categorical (Probability of Detection (POD), False Alarm Rate (FAR), Hanssen-Kuipers index (HK), Frequency Bias Index (FBI)).
  • Precipitation intensity categories: no precipitation (< 1 mm), light precipitation [1–5) mm, moderate precipitation [5–20) mm, heavy precipitation [20–40) mm, and violent precipitation (≥40 mm).

Main Results

• Overall Performance: CHIRPSv3 demonstrated the best overall performance and highest inter-scale consistency, followed by CHIRPSv2 and PERSIANN–CDR.
• Temporal Aggregation: All GPPs showed improved performance with increasing temporal aggregation (daily to 3-day, monthly, and annual scales). CHIRPS products consistently exhibited lower RMSE, higher Pearson correlation coefficients, PBIAS values near zero, and the highest KGE values across all scales. For example, at the daily scale, CHIRPSv3 had an average RMSE of 9.24 mm, and CHIRPSv2 had 9.96 mm, while ERA5-Land and MSWEPv2.8 showed higher errors (10.78 mm and 12.18 mm, respectively).
• Topographic Gradients: GPP performance generally degraded with increasing elevation, particularly for ERA5-Land and MSWEPv2.8. CHIRPS products, however, maintained acceptable-to-optimal KGE values (>0.40) even at higher elevations, with their performance showing an increase with altitude. ERA5-Land exhibited predominantly negative NSE and KGE values in elevated sectors.
• Precipitation Intensity: The detection of daily extreme events remained limited for all products, with moderate improvements observed when considering 3-day accumulations. CHIRPSv2 (POD 0.71) and CHIRPSv3 (POD 0.62) were best at identifying non-precipitation events (<1 mm), while ERA5-Land showed the highest POD (0.64) for moderate precipitation (5-20 mm). All products had very low POD values for violent precipitation (≥40 mm).

Contributions

• Provides the first comprehensive evaluation of CHIRPSv3 performance in a complex tropical region characterized by significant climatic and topographic gradients.
• Offers novel insights into the applicability of GPPs in semi-arid tropical regions with complex topography and limited in situ monitoring, addressing a gap in existing literature.
• Delivers methodological insights and transferable approaches for water resource management in poorly instrumented basins.
• Emphasizes the critical importance of selecting satellite products based on local conditions, such as topography, time scale, and specific management objectives, for reliable decision-making.

Funding

• University of La Guajira, Colombia, under project number CJI-2023–22, as part of the 2023 internal call for Young Researchers.

Citation

@article{Montiel2026Evaluation,
  author = {Montiel, Jhonny I. Pérez and Epiayu, Adriana M. Choles and Moscote, Cristian Diaz},
  title = {Evaluation of satellite precipitation products across climatic and topographic gradients in a basin in Northern South America},
  journal = {Environmental Challenges},
  year = {2026},
  doi = {10.1016/j.envc.2026.101426},
  url = {https://doi.org/10.1016/j.envc.2026.101426}
}