HUANG Chun-Yan, WANG Deng-Wei, XIAO Li-Juan, WANG Ya-Fang
Hyperspectral reflectance data were recorded by the ASD portable non-imaging hyperspectral spectrometer at five growth stages of cotton in an experimental field including two cotton cultivars, Xinluzao No.33, No.13, with four water treatments and cotton canopy thermal images were taken with a Fluke radiometric infrared video camera, thermal images can be used to extract canopy sunlit leaves temperature(TL). Meanwhile, hyperspectral data were processed to calculate the normalized difference vegetation index(NDVI), ratio vegetative index（RVI）, reflectance at red spectrum 620 nm and near-infrared 850 nm wavelength(ρ620, ρ850), respectively. The results showed that canopy leaves temperature(TL) of two cotton cultivars with four water treatments reached the higher value at full flowering stage, full flowering and boll-forming stage, their maximum value were both at full boll stage, the lower TL both occurred at flowering stage and initial open boll stage. Cotton reflectance were reduced at near-infrared and increased at red spectrum wavelength, corresponding NDVI and RVI were less, TL was greater in water stressed plots. The spectrum reflectance at red and NIR wavelength, NDVI, RVI and TL were opposite variation at full water plots than these parameters at water stress treatments. The relationships between NDVI, RVI and TL（rNDVI-TL= -0.4518**，rRVI-TL= -0.5274**，n=38, α=1%）were both stronger than the relationships between ρ620, ρ850 and TL, respectively. It showed that combination of infrared thermography and hyperspectral remote sensing technology have potential to rapidly and nondestructively monitor water status of cotton.
