Monitoramento da seca meteorológica usando dados de precipitação estimados de alta resolução espacial e de longo prazo

Abstract

Drought is a natural disaster that causes water insecurity, especially in semiarid regions, as these areas are vulnerable to the occurrence of this phenomenon, as is the case in part of the Northeast Region of Brazil. Drought analysis requires a set of precipitation data with spatial and temporal precision, however, these data are not always available in sufficient quantity and quality. To monitor the occurrence of meteorological drought, three long-term monthly satellite precipitation data sets called Climate Hazards Group Infrared Precipitation with Stations data (CHIRPS), Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks - Climate Data Record (PERSIANN- CDR), and Climate Forecast System Reanalysis (CFSR) were evaluated and compared with in situ measurements of 38 pluviometric stations for the period 1994-2017 for the Piranhas River basin, Paraíba – Brazil. Statistical indexes were used to assess between observed and estimated data: Determination Coefficient (R²), Relative Bias (BIAS), Efficiency Coefficient (COE), Average Error (ME), and Root Mean Square Error (RMSE). The Standardized Precipitation Index (SPI) was used to estimate meteorological droughts. The obtained results showed a superiority of the values of the CHIRPS and the PERSIANN-CDR in relation to the CFSR. CHIRPS was slightly more accurate on a monthly scale and in spatial distribution when compared to other data sets. Finally, the study points out that the CHIRPS and PERSIANN-CDR data as good alternatives to analyze the spatiotemporal distribution of precipitation in the Piranhas River basin. SPI-12 and SPI-24 satisfactorily detected the main droughts that occurred in the evaluated period. As for satellite data, CHIRPS and PERSIANN-CDR were superior to CFSR in representing these events in the Piranhas River basin.