Space and Time Resolution of a Rain-gauge Network: An Underreported Source of Uncertainty in Extreme Precipitation Estimates

The spatiotemporal resolution of rain-gauge networks fundamentally constrains the reliability of precipitation depth–duration–frequency estimates (DDFs), yet the uncertainty stemming from such restrictions remains unquantified in practice. We evaluated these constraints for short-duration extreme precipitation, using 1-min data from a high-density German network (277 stations) as the reference dataset. We constructed ensembles of lower-resolution networks by progressively reducing station density and aggregating rainfall to 15-min intervals, simulating typical operational networks. Uncertainty was propagated using hierarchical bootstrapping and empirical Bayesian mixture frameworks, closely aligning with U.S. NOAA’s Atlas 14 methods. Reduced station density emerges as a dominant control, inflating DDF uncertainty by up to ~60% with respect to the reference dataset, even for weather-station networks that are still denser than those operated by many national weather agencies. For average recurrence intervals (ARIs) ≥ 5 years, uncertainty scales approximately linearly with gauge density, increasing by ~9 to 20% for each station removed per 1000 km2. Network sparsity also introduces a systematic, ARI-dependent bias–positive for frequent events, negative for rarer extremes–while temporal aggregation causes a consistent negative bias that mean sampling-adjustment corrections cannot fully resolve. Our results are probably an optimistic lower bound in the case of regions characterized by more intense convective activity and steeper spatial gradients in precipitation than Germany (e.g., most of the U.S. or the tropics). These findings suggest that currently reported operational DDF uncertainties are underestimated because they overlook the effects of observational sparsity; moreover, statistical sophistication cannot compensate for such limitations.