Impacts of Spatial and Temporal Station Availability on Gridded Precipitation Products in Central America

Gridded precipitation data sets have become essential for understanding climate variability and long-term trends; however, their accuracy and reliability strongly depend on the availability and the spatial and temporal distribution of in situ meteorological observations. Here, we evaluate the performance of four gridded precipitation products: the Climate Hazards Group InfraRed Precipitation with Station Data (CHIRPS) v2, the Global Precipitation Climatology Centre (GPCC) Full Data Monthly Product v2022, the Climatic Research Unit (CRU) TS 4.07, and the ERA5-Land (ERA5-L) reanalysis, against a network of weather stations across Central America compiled from the regional meteorological service agencies. Using a point (station)-to-pixel comparison and a grid-by-grid spatial decorrelation analysis, we assess gridded data set accuracy and examine how station coverage affects precipitation trend detection. Results from the point (station)-to-pixel analysis show that CHIRPS consistently outperforms ERA5-Land, GPCC, and CRU across all standard statistical metrics (including correlation coefficient, bias, and root mean square error). CRU exhibits the largest spatial decorrelation distances, suggesting inflated spatial coherence likely resulting from interpolation over data-sparse regions. We find disagreement between the spatial representation of precipitation trends between reanalysis-based and observation-based data sets and show that the observed regional drying trend in eastern Honduras and Nicaragua in the GPCC and CHIRPS products may reflect the influence of one station rather than a broader, spatially coherent climate signal. These findings highlight the importance of considering both spatial station density and temporal data availability when using gridded precipitation products for studies of climate variability and change, especially in data-sparse regions such as Central America.