Temporal inhomogeneities in high-resolution gridded precipitation products for the southeastern United States

Abstract. High-resolution gridded precipitation products are widely used in hydroclimatic analyses, although their long-term stability has not been thoroughly evaluated. This study investigates temporal inhomogeneities in five widely used precipitation datasets – Daymet, gridMET, nClimGrid, PRISM, and TerraClimate – across the southeastern United States during 1980–2024. Annual precipitation totals were derived from both monthly and daily data and compared with a reference time series constructed from 120 U.S. Cooperative Observer Program (COOP) gauges. Residual-mass curves and Mann–Whitney U tests were applied to identify temporal inhomogeneities, and trend magnitudes were estimated using the Kendall–Theil robust line. Significant inhomogeneities were detected in most datasets, with nearly 80 % of discontinuities concentrated between 2002 and 2018. These shifts corresponded closely to changes in gauge-network composition and data-processing procedures. Daymet and PRISM exhibited wetting biases linked to the expansion of the Community Collaborative Rain, Hail, and Snow (CoCoRaHS) network and the concurrent decline of COOP gauges, whereas nClimGrid showed a drying bias resulting from increased reliance on Automated Surface Observing System tipping-bucket gauges, which underestimate rainfall. Step increases in TerraClimate and gridMET totals reflected transitions in input data and reprocessing of precipitation forcing fields. These inhomogeneities produced disparate multi-decadal trends ranging from 19 to 48 mm dec⁻¹ compared with a non-significant reference trend of 30 mm dec⁻¹. Among all datasets and combinations tested, the Daymet–nClimGrid pair was the only one without detectable discontinuities and reproduced the reference trend most accurately. This combination provides a homogeneous, temporally consistent dataset for multi-decadal precipitation analyses across the Southeast. Overall, the results demonstrate that unrecognized inhomogeneities in gridded precipitation products can substantially bias regional trend assessments and underscore the need to evaluate and, when necessary, combine datasets to ensure temporal stability in long-term hydroclimatic studies.