Análise de séries temporais cruzadas (NDVI e CHIRPS) na avaliação da memória hídrica e produtividade em áreas terraceadas

Abstract

Increasing climate instability and prolonged water deficits have compromised the hydrological cycle and the resilience of catchment basins, leading to severe urban supply crises. This study aimed to evaluate the eco-hydrological impact of implementing terracing systems in a strategic recharge area within the Alto Graipu micro-basin, located between the municipalities of Guanhães and Sabinópolis (MG). To achieve this, time-series analyses of precipitation (CHIRPS product) and the Normalized Difference Vegetation Index (NDVI, MODIS sensor) corresponding to the period from 2001 to 2025 were employed. Processing was conducted via Google Earth Engine, applying the Savitzky-Golay algorithm for the mitigation of atmospheric noise. The net primary productivity of the pasture (Brachiaria) was quantified using the Area Under the Curve (AUC), and eco-hydrological trends were evaluated through the non-parametric Mann-Kendall test and time-lag analysis. The results showed that conservationist practices installed from 2013 onwards systematically reduced episodes of low vegetative vigor (NDVI < 0.6) during drought periods. A significant increase in AUC and an expansion of the ecosystem's "hydrological memory" for up to 60 days (Lags 1 and 2) were observed, indicating that the terraces acted as slow-release subsurface reservoirs. It is concluded that the micro-topographical modification optimized the infiltration rate, raised the pasture to a new state of productive equilibrium, and mitigated the system's vulnerability to climate fluctuations, strengthening the water security necessary for the municipal public supply.