基于无人机多光谱遥感估算西北半湿润区葡萄基础作物系数研究

To improve the estimation accuracy of vineyard evapotranspiration in northwestern semi-humid area, this study calculated the actual crop evapotranspiration ET_c by the Bowen Ratio System and the reference crop evapotranspiration ET_o based on Penman Formula. The grape crop coefficient K_c was obtained by the division of the two. The FAO-56 double crop coefficient method was used to calculate soil evaporation coefficient K_e and water stress coefficient K_s and obtain the basal crop coefficient K_cb. The spectral data of grape were obtained by using UAV multi-spectral remote sensing image. Reflectance of multiple bands was extracted to calculate four vegetation indexes (Normalized difference vegetation index NDVI, soil adjusted vegetation index SAVI, ratio vegetation index RVI, and difference vegetation index DVI). The relationship model (unary linear regression, polynomial regression and multiple linear regression) between the coefficient of K_cb and vegetation index was established, so as to calculate the actual evapotranspiration of vineyard to verify the accuracy of UAV multi-spectral remote sensing estimation of grape K_cb. The results showed that (1) Under the same modeling method, the model fitting accuracy of vegetation index and K_cb was affected by the species and grape growth period. In the early stage of growth, the fitting accuracy of K_cb -VI_s model obtained by unitary linear regression modeling was NDVI>RVI>SAVI>DVI. In the later growth period, the fitting accuracy was RVI>DVI>SAVI>NDVI. In the whole growth stage, the fitting accuracy was SAVI>NDVI> DVI>RVI. The fitting accuracy of K_cb differed with modeling methods, and the fitting effect of multiple linear regression model was the best. (2) Growth stage, vegetation index type and modeling method were three important factors affecting the accuracy of evapotranspiration estimation. In the early growth stage, the accuracy of the polynomial regression model established by DVI and K_cb was the highest (EF= 0.79). In the later growth stage, the accuracy of the multiple linear regression model was the highest (ET = 0. 80). In the whole growth stage, the validation accuracy of the unitary linear regression model based on DVI and K_cb was the highest (EF= 0.73). (3) The relationship model between K_cb and vegetation index was established at different growth stages. Compared with the K_cb value recommended by FAO-56 double crop coefficient method (EF= 0.58), the inversion K_cb value improved the estimation accuracy of evapotranspiration by more than 6%.