NONLINEAR CLIMATE-VEGETATION INTERACTION IN SUGARCANE: A CASE STUDY IN JUAZEIRO, BAHIA
Remote sensing; Quantile regression; Vegetation indices; Semi-arid agriculture; Land–atmosphere interactions.
Sugarcane production in semi-arid environments is strongly conditioned by land–atmosphere interactions, modulated by radiation, precipitation, and vegetation dynamics. In this context, this study investigates the non-linear and quantile-dependent responses of mean relative humidity (RH) to global solar radiation, precipitation, and vegetation variability in a commercial sugarcane system located in the semi-arid region of Juazeiro, Bahia, Brazil. Daily meteorological data were obtained from an automatic agrometeorological station operated by the Brazilian Semi-Arid Agricultural Research Center and subsequently aggregated to a biweekly scale. Vegetation dynamics were characterized using spectral information derived from the MODIS sensor aboard the Terra satellite, employing the NDVI and EVI indices extracted from the SATVeg platform developed by Embrapa Digital Agriculture. A non-linear quantile regression approach was applied at the biweekly scale to capture asymmetric and threshold-dependent relationships across the RH distribution. The results indicate that precipitation exerts a dominant positive control on the upper RH quantiles, whereas global solar radiation acts as a consistent negative modulator, particularly under drier atmospheric conditions. EVI exhibited higher sensitivity and more pronounced non-linear responses under regimes of high humidity and precipitation, while NDVI showed more stable and near-linear relationships across the median and upper quantiles. These findings highlight the combined role of vegetation physiological activity and surface energy balance in regulating atmospheric moisture and demonstrate the potential of integrating agrometeorological data and remote sensing for eco-hydrometeorological monitoring of climate-sensitive agricultural systems in semi-arid regions.