Unraveling Crop Nitrogen-Water Dynamics with Hyperspectral-Thermal Sensing in Northern Central Valley, California

Abstract Ensuring global food security in the face of climate change requires optimizing crop water use and nutrient management. This study investigates the relationship between canopy nitrogen (N) and evapotranspiration (ET) across sunflower, rice, walnut, alfalfa, and plum crops using advanced remote sensing technologies. High-resolution hyperspectral data from NASAs Earth Surface Mineral Dust Source Investigation (EMIT) and thermal multispectral data from the Landsat-based OpenET system were analyzed over 1,135 km 2 in California. Regression analysis revealed strong spatial association between canopy N and ET for sunflower (R 2 = 0.82), rice (R 2 = 0.71), and walnut (R 2 = 0.68), with weaker associations for alfalfa (R 2 = 0.38) and plums (R 2 = 0.26) over the average of growing season. Temporal variations highlighted the high nitrogen efficiency of sunflower, rice, and walnut, demonstrating the capability of hyperspectral sensors to monitor nutrient-water dynamics effectively. These findings emphasize the potential of integrating hyperspectral and thermal data to enhance fertilizer and irrigation strategies. Furthermore, the study aligns with NASAs Surface Biology and Geology (SBG) mission, offering valuable insights into precision agriculture and sustainable crop management under changing climatic conditions. Graphical Abstract Based on the graphical snapshot, this study was conducted to investigate canopy nitrogen (N) and evapotranspiration (ET) dynamics using hyperspectral and thermal remote sensing in a 1,135 km 2 agricultural region in Californias Northern Central Valley. This work captures the complex relationships between nutrient-water dynamics and agricultural productivity across five crops: sunflower, rice, walnut, alfalfa, and plums, mapped using USDA NASS Cropland Data Layers. Data collection utilized hyperspectral data from the EMIT satellite and thermal data from Landsat-based OpenET, focusing on spatial and temporal variations across four dates (June 11, June 26, August 3, 2023, and January 11, 2024). Analytical methods included validation of canopy N estimation through vegetation indices (NDVI, NDNI, MCARI) derived from EMIT data, supported by scatter plots, boxplots, and statistical analyses such as ANOVA, alongside crop-specific analysis of the N-ET relationship. The Partial Least Squares Regression (PLSR) model was applied to EMIT hyperspectral data to generate temporal N maps, revealing underlying nutrient patterns. Results demonstrated strong N-ET spatial associations over the average of growing season for sunflower (R 2 = 0.82), rice (R 2 = 0.71), and walnut (R 2 = 0.68), with weaker associations for alfalfa (R 2 = 0.38) and plums (R 2 = 0.26), while vegetation indices showed high predictive accuracy (R 2 up to 0.95 for rice and sunflower). Temporal N, ET, and nitrogen-to-ET ratio (NER) maps highlighted efficient N use in low ET areas during January. The findings underscore the utility of integrating hyperspectral and thermal remote sensing to track nutrient-water dynamics, offering critical insights for developing crop-specific, seasonally adjusted management strategies to optimize resource use in precision agriculture.