Erosividade da chuva na América do Sul: comportamento atual e projeções futuras

Abstract

The intense precipitations and their erosive potential have been studied on a continental scale due to their importance for the planning of the environment as well as for agriculture, which can negatively affect the economy and society in a significant way. However, the erosive potential of rainfall, known as rainfall erosivity, and the mechanisms that control it, are complex studies on a large scale, especially in regions with limited rainfall data observed by rain gauges. This research has as main objective the mapping the annual mean R factor of the Universal Soil Loss Equation, revised version (RUSLE) for South America for the present time and throughout the 21st century, identifying regions that are more vulnerable considering their natural variability based on existing rainfall erosivity records (El30) and their respective geodesic, altitude and mean annual precipitation for South America, a statistical model was generated to determine the mean R factor throughout the continent, using the stepwise method for selection of input variables. For the current behavior was used WorldClim data for quantify the future behavior of the R factor was used daily average precipitation data based on climate change projections of MCG HadGEM2-ES, MIROC5 and CanESM2 regionalized by Eta/CPTEC model for RCPs 4.5 and 8.5 for the period 2010-2099, subdivided by period (2010-2040; 2041-2070; 2071-2099). The generated statistical model presents good performance with coefficient of determination superior to 0.60. By means of techniques available in ArcGIS software, version 10.1 (ESRI, 2014), a map of the factor R was subdivided into 18 classes, with spatial resolution of 90m. Thus, it was possible to map the subdividing factor R in 18 classes, and spatial resolution of 90m, with spatial gradient of the R factor moving in the southwest-northeast direction. This map presents variation between 50 and 23000 MJ mm (ha h)−1 year−1, for the present time, with low erosive potential in RAC and high in RN, in the order of 139 and 6761 MJ mm (ha h)−1 year−1, being a direct reflection of the spatial distribution of precipitation in the continent. Projections of the factor R throughout the 21st century by simulations of MCG show agreement for significant increases for RAC (500%), accentuating to RCP8.5; for RAS, factor R increase (290%) is also projected; for RN there is a decrease in the order of 47% and 83% (RCP4.5 and 8.5) for the period 2071-2099; for the RAN and RNE, a reduction trend in the order of 56% and 91% (RCPs 4.5 and 8.5, period 2071-2099) is observed, however, the Eta-MIROC5 model simulates R factor elevation for RAN and RNE (15% and 45%, RCPs 4.5 and 8.5), characterizing uncertainties regarding factor R in these regions. Finally, for RS and RCS, a higher increase (up to 120%) was observed for RCP 8.5 (period 2071-2099), however for the other periods there is projection of reduction.