Thermal imaging and passive reflectance sensing to estimate the water status and grain yield of wheat under different irrigation regimes

The water demand for agricultural purposes is steadily increasing. The use of contactless sensing techniques, such as passive reflectance sensors and thermal imaging cameras, is therefore becoming imperative and will be one of the major adaptation strategies to control the irrigation schedule under arid and semi-arid conditions. In this study, the performance of hyperspectral passive reflectance sensing and infrared thermal imaging was tested to assess their relationship with the water status and grain yield (GY) of wheat cultivars via simple linear regression and partial least square regression (PLSR) analyses. The models included data of the (i) normalized relative canopy temperature (NRCT); (ii) PLSR based on selected spectral indices; (iii) data fusion model of PLSR based on selected spectral indices and the NRCT; and (iv) data fusion model of PLSR based on selected spectral indices, NRCT, relative water content (RWC), and canopy water content (CWC). The experimental treatments involved two wheat cultivars (Gmiza 11 and Sods 1) and three water regimes (irrigated with 100%, 75%, and 50% of estimated crop evapotranspiration). The results show that the NRCT was closely and significantly associated with RWC, CWC, and GY, with R2 = 0.84, 0.87 and 0.81, respectively. The data fusion model of PLSR based on selected spectral indices, NRCT, RWC, and CWC improved the yield prediction under three irrigation regimes (R2 = 0.97, slope = 0.99, root-mean-square error = 26.48 g/m2). In conclusion, improvements can be made in the yield prediction when traits that are physiologically related in different ways to the yield are combined with non-destructive data.