Análise da distribuição espaço temporal de variáveis hidrológicas e biofísicas no semiárido de Pernambuco

Abstract

The Brazilian semi-arid region has limited availability of water resources. In addition, profound changes in land use and occupation have occurred in watersheds of Pernambuco State, such as the implementation of the São Francisco River Integration Project (PISF). This dissertation aims to analyze the spatial distribution of hydrological variables in the semi-arid region of Pernambuco in two basins of this region, one that already has the PISF water supply and while the other does not, but there are future projects for its implementation through branch, are the basins of the Terra Nova and Brígida rivers, respectively. In the Terra Nova river basin, the objective was to evaluate the real evapotranspiration and map cultivated areas through remote sensing in a perennial stretch. Landsat-8 satellite images from 2015 to 2020 were selected. The images were processed in Google Earth Engine (GEE) and edited in QGIS 3.16 software. It was noticed by the NDVI, an increase in the vegetation cover index spatially. Regions with higher values of real evapotranspiration are linked to those with lower temperatures. A smaller amount of cultivated areas was observed in the Terra Nova River stretch in the 2015 images and the expansion of agriculture in the region on the banks of this river, in its perennial stretch. In addition to the pluviometric regime, the release of water from the PISF contributed to the increase in irrigated areas in the region. And in the Brígida river basin, the objective was to evaluate the distribution of precipitation in years under different rainfall regimes and its impacts on the dynamics of vegetation cover, addressing the water-vegetation nexus for the region, where water supply is strongly dependent on the occurrence of rains. Rainfall data from the Climate Hazards Group InfraRed Precipitation with Station (CHIRPS) was used in conjunction with observed data from 40 weather stations with an annual time series of 55 years (1962 to 2017). The Standardized Precipitation Index (SPI) was applied to assess the annual variability of precipitation (SPI-12). The data were submitted to classical statistics and geostatistical analysis, using ordinary kriging (KO) and Sequential Gaussian Simulation (SGS) methods to map the spatial distribution of rainfall. A comprehensive set of fifteen remote sensing images for years with different rainfall regimes was analyzed, allowing the calculation of the Normalized Difference Vegetation Index (NDVI) and the Modified Normalized Difference Water Index (MNDWI). There was a high correlation between observed and estimated data. The geostatistical analyzes showed models with strong spatial dependence for all the years adopted. Through kriging maps, it was found that the rainfall in the basin increases strongly with altitude. When the number of SGS realizations increases, SPI maps tend to be stable and capture inherent variability not represented by the kriging procedure. Among the established and validated semivariograms, the spherical model was the one that best fitted the data set used. Through the SGS technique, it was verified the low uncertainty of the CHIRPS data and also that 100 realizations are enough to generate SGS maps suitable for the SPI in the basin. Even in normal years, the number of water bodies is low, which can compromise water security in the region.