In this study we analyze the relationship between the occurrence of high temperatures and the variability of soybean yield. The study covers the crop years from 2003 to 2012 in Rio Grande do Sul State. For the analysis of temperature over the vegetation canopy in agricultural areas, we used a LST (Land Surface Temperature ) time series, product MYD11A2 collection -5 of sensor MODIS ( Moderate -Resolution Imaging Spectroradiometer ) on board the Aqua satellite. Crop area mapping, identification and monitoring of development during phenological stages were performed with application of EVI (Enhanced Vegetation Index) data series product MOD13Q1, collection-5 of MODIS on board the Terra satellite. Ordinary kriging methodology was used for data smoothing of spatial distribution of soybean yield at municipal level, obtained from IBGE; accumulated rainfall, between October and February, came from data of 15 meteorological stations from FEPAGRO (Fundação Estadual de Pesquisa Agropecuária). The overall average LST over crop canopy in the studied period was 31.9 ºC with standard deviation 2.1 °C. The overall average of the smoothed yield was 2003 kg ha -1 with standard deviation 309 kg ha -1. The average of cumulative rainfall was 775 mm with a standard deviation of 38 mm. At the state level analysis, results indicate that severe decreases in productivity are coupled events effect of canopy-LST above average during February (DOY 033), associated to below average rainfall by means an inverse nonlinear relationship with R2 = 0.73 and RMSD = 17,8%.

Revista Brasileira de Geografia Física