Estimating Crop Coefficient Of Field Maize Based On Uav Multispectral Remote Sensing

The rapid acquisition of crop coefficient K_c is the key to the estimation of field crop evapotranspiration.Based on field maize,soil and meteorological data at the experimental station in Zhaojun Town,Dalate Banner,Inner Mongolia,2017,this paper used the dual crop coefficient method calibrated by meteorological factors and crop coverage to calculate maize crop coefficient of different growth stages and different water stress.using self-developed UAV multi-spectral system to take corn canopy multi-spectrum?Blue,Green,Red,Red Edge,Near IR,475⁸40nm?images,using different modeling methods to study the relationship between six different types of vegetation indices?normalized difference vegetation index,NDVI,soil adjusted vegetation index?SAVI?,enhanced vegetation index?EVI?,ratio vegetation index?SR?,greenness normalized vegetation index?GNDVI?,resistance index?VARI??and crop coefficient K_c and basic crop coefficient K_cb in different growth stages?different crop growth periods in FAO56:rapid growth stage,middle growth stage,and late growth stage?and researching the effect of water stress on the crop coefficient.Generating the spatial distribution map of K_c and actual evapotranspiration ET and analyzing the feasibility and applicability of UAV multi-spectral remote sensing technology to estimate crop coefficient K_c.The main work and conclusions of this paper are as follows:?1?Using the self-developed UAV multi-spectral system to collect multi-spectral images of corn in the experimental area,Pix4DMapper software was used to stitch 1795high-resolution multispectral images acquired by UAV multispectral systems to obtain orthographic images of the entire experimental site.The ground resolution was 0.05m.The orthophoto image was obtained after the grey plate correction and the reflectance images of each band were obtained.?2?The accuracy of the relationship between corn vegetation index and crop coefficient K_c was established by the support vector regression method under full irrigation from rapid growth period to late growth period is the best,and the precision of vegetation index SR and crop coefficient model?R²=0.64,RMSE=0.0526?is the best,but the model accuracy is relatively poor;the accuracy of model between vegetation index and crop coefficient K_c established under water stress conditions using multiple linear regression is the best?R²=0.90,RMSE=0.0996?,but the model deviation is relatively large.A linear regression method was used to establish the model between vegetation index and basic crop coefficient K_cb in the period of rapid growth to late growth stage under adequate irrigation conditions.The accuracy of model between vegetation index EVI and K_cb?R²=0.83,RMSE=0.0834?is the best;Under the stress condition,a linear regression method was used to establish the model between vegetation index and basic crop coefficient K_cb,the model precision of vegetation index SAVI and K_cb?R²=0.82,RMSE=0.0857?is the best.?3?Growth period and water stress are the two main factors affecting the crop coefficient K_c estimated by the vegetation index of maize.The correlations between vegetation indices and K_c of different growth stages and different water stress maize are quite different.In the fast growing period,a linear regression method was used to establish the model between vegetation index and crop coefficient K_c,the model precision of crop index SR and crop coefficient?R²=0.92,RMSE=0.0296?is the best;under the water stress conditions using the multiple linear regression method to establish the model between vegetation index and crop coefficient K_c and model accuracy is the best?R²=0.30,RMSE=0.0327?.The accuracy of model between vegetation index and the crop coefficient K_c using the support vector regression method is the best under the condition of full irrigation from middle growth to late growth stage,and the precision of model between the vegetation index SR and crop coefficient K_c is the best?R²=0.44,RMSE=0.0789?.Under the condition of water stress support vector regression was used to establish the model between vegetation index and crop coefficient K_c,and the accuracy of model between index of vegetation index SR and crop coefficient K_c is the best?R²=0.94,RMSE=0.0728?.?4?Using linear regression method to establish the model between the vegetation index SR and crop coefficient K_c in the rapid growth stage and generating the sample A spatial distribution map of crop coefficient K_c and actual evapotranspiration ET.From the middle growth stage to the late growth stage,using polynomial regression method to establish the model of the vegetation index SR and crop coefficient K_c and generating the sample B spatial distribution map of crop coefficient K_c actual evapotranspiration ET.The results showed that the true value of the actual value of the actual evapotranspiration in the fast growing period was estimated?R²=0.83,RMSE=0.1909?and the accuracy of the actual value estimated by using the actual evapotranspiration from the late growth period to the late growth period?R²=0.87,RMSE=0.5542?is better.