Spatial pattern probabilities exceeding critical threshold of annual mean storm-erosivity in euro-mediterranean areas

Abstract

In contrast to the moderate amounts of yearly average rainfall, the recurrence of heavy rainstorm can be considered a critical hydro-climatological feature for land-and-soil conservation-and-planning of the river basins. This work illustrates a spatial modelling study of rainstorm aggressiveness to assess downscaling in the erosive rainfall climatic classification across Euro-Mediterranean regions. Rainfall erosivity was estimated by the R–climatic factor of the RUSLE approach at 102 raingauges across Europe. For this purpose, an issue model of kriging, termed as lognormal probability cokriging (LPCK), is emphasized to a soft description of the erosive hazard in terms of probability, which is consistent to mitigate the uncertainty of the rainfall erosivity spatial classification. For improving spatial prediction, multivariate geostatistical modelling uses the rainfall 95th percentile at about a 1,000,000 of grid-points as auxiliary information, when the erosivity information is transferred from point to landscape. The estimate of uncertainty at unsampled raingauge via LPCK, was used to explain the probability of exceeding the thresholds of 1,000 and 1,500 MJ mm ha-1 h-1 year-1 as critical values classes of the erosivity at a spatial resolution around to 25 km. In this way, about the 50 % of the area has a probability higher of the 70 % subjected to a rate exceeding to 1,000 MJ mm ha-1 h-1 year-1. The area hit by storm erosivity with a rate higher of 1,500 MJ mm ha-1 h-1 year-1, drops to around 10 %, although, with a probability of 50 %, the surface remains still large (about 40 % of the area).