INTRODUCTION
The Mediterranean Iberian area is prone to extreme precipitation events due to the Mediterranean atmospheric dynamics, which can generate high-intensity convective storms. This was demonstrated in October 2024, when an extreme cut-off low impacted the eastern coast of Spain, triggering intense convective instability and exceptional rainfall accumulations.
This research assesses spatial interpolation methods to generate rainfall fields from observations at rainfall-gauging stations during localized convective storms to support forecasting and operational workflows through improved precipitation representation.
METHODOLOGY
This research considers the October 2024 flood event in the El Poyo Ravine catchment as a case study, covering 28 October 2024 at 07:00 to 30 October 2024 at 06:45 including the intense phases of the event. The study uses 15-min rainfall observations at rain-gauging stations from two networks.
Eight interpolation methods were considered: Thiessen polygon (TP), inverse distance weighting (IDW), local polynomial interpolation (LPI), trend surface interpolation (TSI), ordinary kriging (OK) (both local and global), cokriging (CK) with altitude as a secondary variable, and a mathematical formulation-based method (CONV).
Validation followed a hierarchical scheme, prioritizing root mean square error (RMSE) and Nash–Sutcliffe efficiency (NSE), followed by mean absolute error (MAE) and mean error (ME), to select the optimal parameter configuration within each method and, subsequently, the best-performing interpolation method.
RESULTS
Overall, most deterministic methods considered provide accurate estimates at relatively low computational cost, comparable to equation-based approaches. By contrast, geostatistical methods require greater computational effort. Runtime was mainly influenced by the number of rainfall-gauging stations.
While kriging is generally less suitable for convective storms, global ordinary kriging yielded the best results in this case.
CONCLUSIONS
Although limited to one catchment and extreme event, this study provides a transferable methodology for similar Mediterranean catchments and convective events. Under different conditions, the methodology should be replicated to identify the optimal interpolation method.