Inverting the water stress index of the Brassica chinensis using multiple-spectral and meteorological parameters

Abstract: Monitoring the Crop Water Stress Index (CWSI) is of great significance for the water status and irrigation in crop production. Taking the Brassica chinensis as the test object, this study aims to measure the canopy temperature under different soil moisture conditions. Some meteorological parameters were collected, including the air temperature, relative humidity, wind speed, and photosynthetic active radiation. Meanwhile, the images of spectral reflectance were also collected for the four bands (450, 650, 808, and 940nm). Four vegetation indexes were then calculated by the canopy spectral reflectance, including the Normalized Difference Vegetation Index (NDVI), Difference Vegetation Index (DVI), Re-Difference Vegetation Index (RDVI), and Optimized Soil-Adjusted Vegetation Index (OSAVI). Support Vector Regression (SVR) was selected to construct the inversion models of the CWSI upper/lower baseline using the meteorological parameters, and the inversion models of the canopy temperature using the vegetation index. The results showed that the canopy spectral reflectance at 450 and 650 nm for the Brassica chinensis ranged from 0 to 0.1, while the relatively higher one at 808 and 940 nm ranged from 0.4 to 0.6. The reflectance at 808 and 940 nm increased outstandingly, when the Brassica chinensis was developed gradually from the vegetative to reproductive growth stage. The vegetation index reflected the growth state and vegetation coverage of the Brassica chinensis. There was a different response of vegetation indexes to the canopy temperature. The vegetation NDVI, DVI and RDVI increased, while the vegetation OSAVI decreased with the increase of the canopy temperature of the Brassica chinensis. The vegetation index under the same water treatment was slightly different in the various growth stages. Specifically, the range of the vegetation index in the reproductive growth stage was smaller than that in the vegetative growth stage. The error analysis showed that the inversion models were feasible to monitor the air temperature, relative humidity, wind speed, and photosynthetic radiation, further invert the upper/lower baseline of CWSI with the determination coefficient greater than 0.75. In the light of the error analysis of the inversion models, the vegetation index was inverted the canopy temperature of the Brassica chinensis in the vegetative and reproductive growth stage, indicating an excellent accuracy with the determination coefficient greater than 0.7. The calculated CWSI using the inversion models presented a significant correlation with the using the measurement, while the determination coefficient was equal to 0.70. And the CWSI showed the negative relationship with the stomatal conductance with the determination coefficient equal to 0.53. The meteorological parameters were used to invert the upper/lower baseline of CWSI, where the vegetation indexes were used to invert the canopy temperature. The inverted values using the SVR model shared the better fitting performance. The finding can provide a strong support for the spectral monitoring of the crop water stress index of the Brassica chinensis.