Remote sensing drought indices for the semi-arid region of southeast Brazil.

The semi-arid region of the Minas Gerais state, Southeast Brazil, has experienced some events of drought, while arising large irrigation schemes in some areas increasing water withdrawn from the rivers. These scenarios demand large-scale water balance studies to subsidize water resources policies. The visible, infrared and thermal bands from Landsat 8 satellite images were used together with a net of agrometeorological stations to test drought indices in this region, during the year 2015, classifying irrigated areas and natural vegetation, aiming mitigations for climatically water scarcity. The ETr index (i.e., the ratio of actual ? ET to reference ? ET0 evapotranspiration) was obtained by using the SAFER (Simple Algorithm for Evapotranspiration Retrieving) algorithm, while the SUREAL (Surface Resistance Algorithm) algorithm was applied to estimate the surface resistance to water vapor transfer (rs). The highest and lowest ETr and rs values, meaning better root-zone moisture conditions, above 0.90 and bellow 800 s m-1, respectively, happened in the counties with the largest concentrations of irrigation area, resulting in an increment on ET of 3.0 mm d-1, because of the replacement of natural vegetation by irrigated crops. On the other hand, outside the rainy period, natural vegetation showed ETr values close to 0.00 and rs very high, above 1000 s m-1. Between the two drought indices, rs detected better the different soil moisture conditions. The results are relevant for monitoring drought events under the additional environmental consequences of land use change in semi-arid environments