ANALYSIS OF INFLUENCE OF EVAPOTRANSPIRATION ON RAINFALL IN AN ATLANTIC FOREST USING REMOTE SENSING DATA

Abstract

Quantifying and monitoring evapotranspiration in large natural forests is vital for monitoring the climate, water resources and biodiversity of the ecosystem, especially in an area with high climatic variability, water and biodiversity, such as Northeast Brazil. The objective of this study is to evaluate evapotranspiration in areas of the Atlantic Forest using remote sensing data and to determine how evapotranspiration can affect rainfall changes on a local scale. We used the Surface Energy Balance Algorithm for Land (SEBAL) and Thornthwaite models to estimate evapotranspiration. The input models used Landsat images and observations from agro-meteorological stations. An estimate of the actual evapotranspiration by the Thornthwaite method was over a 10-day period, and SEBAL was used over 24 hours. The daily actual evapotranspiration was between 5 and 6 mm/day. We computed the cross correlation coefficient and standard error with lag seven (10-day) for maximum evapotranspiration with SEBAL (Evap_SEBAL_max) and rainfall decennial (R10). The evapotranspiration had a positive correlation with rainfall. The correlation is significant for the last ten days (lag-1), with r=0.389. While this value does not indicate a strong statistical correlation by itself, physically it designates a positive influence of the local evapotranspiration on local precipitation, given all the other environmental aspects that influence precipitation. The results suggest that local evapotranspiration can be influencing the local rainfall. Thus, knowing the amount of water being used by the forest and released to the atmosphere is important. These results can be used for improve climate prediction and climate change models at the local scale.