CORRELATIONS BETWEEN CLIMATIC VARIABLES AND MULTITEMPORAL EVI IMAGES IN NATURAL FOREST AND CROPPING AREAS

Abstract

Through the utilization of multitemporal series of EVI (Enhanced Vegetation Index) data and climatic variables, 990 models have been adjusted, divided between natural forest environments and cropping areas. The study area locates in the city of Sooretama, to the north of the State of Espirito Santo, between the parallels 18º 53 53"S and 19º 30' 00"S and the meridians 39º 50 00"W and 40º 15 50"W. It s been assumed that the EVI data vary exclusively in function of the pluviosity (PP), temperature (T), relative humidity (UR) and photoperiod (FP). So, these four variables were grouped in all possible ways, totalizing fifteen combinations. In the investigation of the correlations between EVI and climatic variables, were also considered the lag that the vegetation needs to overwhelm the stress (e.g. hydrical stress). The models have been adjusted through the ordinary least-squares method and evaluated by the determination coefficients (R2) values. It was concluded that while in the natural forest the EVI has a higher correlation with climatic variables applying more days of lag, in cropping areas the inverse was observed. Models contents two variables showed to have a predictive capacity close to the ones with four variables. The models with bigger predictive capacity for the natural forest had included the FP, whereas for the cropping area they had been the ones that had included the T. The 3 independent variables in the cropping areas more correlated with the EVI were T, UR and PP.