From means to extremes: Evaluating high-resolution gridded data for rainfall erosivity estimation in Austria

Rainfall erosivity (EI 30 ) is a key driver of soil erosion and requires high-resolution rainfall data to capture extreme events and their spatiotemporal variability. This study evaluates the potential of high-resolution gridded rainfall data, the Integrated Nowcasting through Comprehensive Analysis (INCA) dataset (15-minute and 1 km*1 km), for estimating EI 30 across Austria's Main Agricultural Production Zones (MAPZ). We assessed (i) the ability of INCA to estimate average annual EI 30 values, (ii) its performance in capturing single extreme events, (iii) its capacity to model extreme erosivity patterns using Generalized Extreme Value (GEV) distributions, and (iv) the impact of temporal resolution (15- and 30-min) on EI 30 estimates, using data (5-minute) from 26 rain gauges. Results indicate that INCA performs well in estimating average annual and maximum annual erosivity, with underestimations of 3.8% and 13.1%, respectively. Spatially, underestimations were most prominent in the southern pre-Alpine regions, while overestimations occurred in the eastern lowlands. The gridded data effectively captured large EI 30 values exceeding 3000 MJ ha⁻¹ h⁻¹ . GEV analysis revealed strong agreement between gridded and rain gauge data for return periods of 2–20 years. However, INCA exhibited significant limitations in detecting specific extreme events. Increasing the temporal resolution from 30 to 15 min improved the estimation of EI 30 , underscoring the value of INCA’s 15-minute temporal resolution for capturing EI 30 patterns. This study highlights the potential of INCA for reproducing EI 30 patterns across MAPZ, particularly in regions with sparse rain gauge networks, while also identifying areas for further refinement in capturing extreme events. • INCA performs well in estimating average annual and maximum annual EI 30 , with underestimations of 3.8% and 13.1%, respectively. • INCA captures large EI 30 event values exceeding 3000 MJ ha⁻¹ h⁻¹ . • Strong agreement between gridded and rain gauge data for return periods of 2–20 years. • Gridded data show significant discrepancies when estimating extremes at the specific-event level. • Temporal aggregation of rainfall data—to 15- and 30‑minute intervals—affects estimates of I 30 , E, and EI 30 .