Multispectral, Thermal, and Hyperspectral Sensing Data Depict Stomatal Conductance in Grapevine

Abstract

Climate-driven water challenges in the Pacific Northwest necessitate precise irrigation for sustainable vineyard management. In such scenarios, conservation of water using different approaches, including subsurface irrigation, becomes critical. Detecting crop water status becomes key to evaluating and managing such approaches. This study examines how multispectral, thermal, and hyperspectral proximal sensing data depict irrigation-induced variations in stomatal conductance in Cabernet Sauvignon vineyards during 2016 and 2017. The roles of individual and combined sensing modalities were analyzed, with key contributions including the identification of indices that characterize stomatal conductance. Data were collected at the following growth stages: 80 and 44 days before harvest (DBH) in 2016; and 64, 44, and 8 DBH in 2017. The vegetation indices analyzed included the green normalized difference vegetation index (GNDVI) and leaf area index (LAI) from multispectral data, crop water stress index (CWSI) from thermal data, and normalized difference spectral indices (NDSI) from hyperspectral data. Pearson’s correlations at 80 and 44 DBH (2016) showed significant relationships between normalized stomatal conductance and multispectral indices (LAI: r = 0.59 to 0.66, GNDVI: r = 0.41 to 0.50, both p r = −0.27, 0.31, both p r = −0.83) at 44 DBH. In the same year, NDSI pairs exhibited stronger correlations than multispectral indices as the DBH decreased (1380 nm with 1570 nm: r = −0.58 to −0.69, 1570 nm with 1810 nm: r = 0.64 to 0.48, both p R2 = 0.37–0.50; 2017: R2 = 0.51–0.63, both p < 0.01). These results demonstrate the precision of a multimodal sensing approach, particularly integrating multispectral and hyperspectral data, to improve irrigation strategies and promote sustainable viticulture.

Keywords

• proximal sensing
• NDSI
• water stress
• vineyard monitoring