Evaluation of evapotranspiration variations according to soil type using multivariate statistical analysis

Abstract

It is known that the soil type and its characteristics are very important to the agriculture and to maintain the health of a forest. Thus, the objective of this work was to verify, through multivariate statistical analysis techniques, the variations of evapotranspiration values as a function of the soil type in three different vegetation cover areas: sugarcane, planted forest and native forest. For this, an area of 2231.926 km2 was chosen, using monthly Landsat 8 satellite imagery with Orbit/Path 220 and Point/Row 75, between latitudes 20°43′55.2″S and 22°36′28.8″S and longitudes 46°37′48,0″W and 48°50′16,8″W, in the region of Corumbataí, northeast of the State of São Paulo, Brazil, collected over 24 months between April 2013 and March 2015. The evapotranspiration evaluation was obtained by satellite imagery using a hybrid model of the Two Source Energy Balance (TSEB), with a proposed adjustment methodology to convert hourly ET to monthly and annually data. The multivariate statistical analysis of monthly ET patterns across soil type classified three evapotranspiration groups in the three studied land cover areas. In the areas with sugarcane, evapotranspiration separated statistical groups in the areas of the Cerrado biome with Ferralsol (LVA) and Lithic Leptosol (RL), another group in the areas with Ferralsol (LV) and Ferralic Arenosol (RQ), and a third group of Acrisol (PVA) in an Atlantic Forest biome area. In the areas with planted forest and native forest, evapotranspiration separated a first group with soils LVA, RQ, PVA, Dystric Gleysol (GX) and RL in areas of Cerrado, a second group with LV in Atlantic Forest area and a third group just with Fibric Histosol (OY).