Research On UAV Remote Sensing Estimation Methods Of Evapotranspiration Of Maize Field

China is one of the countries with the poorest freshwater resources in the world,and the proportion of agricultural water consumption in the national annual water consumption distribution remains high.Improving agricultural water use efficiency is the key to achieve sustainable use of water resources,while field evapotranspiration is an important link in farmland microclimate and hydrological cycle.As an important component of surface water heat balance,field evapotranspiration’s rapid and accurate estimation plays an important role in accurately formulating field irrigation systems and improving field irrigation water use efficiency.At present,both the traditional evapotranspiration calculation method and the satellite remote sensing method are unable to meet the rapid and accurate estimation requirements of daily evapotranspiration of field scale.To explore the feasibility and applicability of unmanned aerial vehicle（UAV）remote sensing estimation of field evapotranspiration,the spatial distribution information of field evapotranspiration can be quickly and accurately obtained to achieve precision field management,this article takes the maize field under different water treatment conditions in Zhaojun Town,Dalate Banner,Ordos City,Inner Mongolia Autonomous Region in 2018 and2019 as the research object.Using the DJI Phantom 4 Pro UAV visible light system,the self-developed UAV multi-spectral system,and UAV thermal infrared system to study the classification methods of the maize canopy and soil surface at different growth stages of the field.On this basis,parameters such as plant height（h）,leaf area index（LAI）,canopy temperature（T_c）,and soil surface temperature(T_soil)were extracted respectively,and the changes of various parameters in different growth stages were analyzed.Finally,three evapotranspiration estimation methods are improved,namely,the dual crop coefficient method,the single-layer energy balance method,and the dual-layer energy balance method.A study on the estimation of evapotranspiration from maize fields based on drone remote sensing technology was carried out.The main research contents and conclusions of this paper are as follows:（1）To study the different classification methods in the different field corn growth stage applicability,to determine the optimal classification method of field canopy and soil surface based on visible light remote sensing image of UAV,this paper analysis the object–oriented-SVM,maximum likelihood,artificial neural network,random forests and K-means methods for different stages of maize growth field canopy and soil surface classification results,and analyzed the variation rule of different classification methods accuracy at the different growth stage of maize.The research shows that due to the change of vegetation coverage,the classification accuracy of field maize canopy and soil surface first decreased and then increased from the rapid growth period to the late growth period,and the classification accuracy of the rapid growth period was the highest;the object-oriented-SVM classification can accurately classify the field canopy and soil surface of maize at different growth stages（Overall accuracy≥96.03%,Kappa coefficient≥0.918）.It provides a technical basis for the rapid and accurate remote sensing estimation of evapotranspiration in the field.（2）Field corn evapotranspiration estimation based on the dual crop coefficient method.To realize the rapid and accurate estimation of field corn evapotranspiration based on the dual crop coefficient method,unmanned aerial vehicle remote sensing was used to extract h and LAI of field corn,and on this basis,the dual-crop coefficient method was improved.The results showed that combining the results of field maize canopy and soil surface classification,using the UAV digital surface model of different maize growth periods and the bare field digital surface model to do the subtraction can accurately extract the field maize plant height information（2018:R²=0.97,RMSE=0.09 m;2019:R²=0.96,RMSE=0.09 m）.Through the support vector regression method,LAI prediction models at different growth stages of maize established in combination with different years and vegetation index were able to predict LAI of field maize accurately（rapid growth period:R²=0.9620,RMSE=0.0440;middle growth period:R²=0.9658,RMSE=0.0908;late-growth period:R²=0.9760,RMSE=0.0644）.The evapotranspiration estimated by the improved dual crop coefficient method is like that calculated by the original dual crop coefficient method（R²≥0.95,RMSE≤0.17 mm）.The improved dual crop coefficient method realizes the rapid and accurate extraction of the spatial distribution information of field corn evapotranspiration.（3）Extraction of maize canopy temperature and soil surface temperature.The influence of UAV flight height on the temperature of the thermal infrared image was studied,aiming at the flight altitude（60 m）of the UAV thermal infrared system,the temperature correction model of thermal infrared images in different periods was established（2018:R²≥0.91;2019:R²≥0.92）,verification shows that the corrected thermal infrared image can accurately extract the surface temperature（2018:R²=0.97,RMSE=0.81℃;2019:R²=0.98,RMSE=0.86℃）.According to this basis,T_c and T_soil were extracted separately based on UAV remote sensing technology,and the changes in surface temperature（T_s）,T_c,and T_soil in different maize growth stages were analyzed.The research results showed that combining the classification results of the maize field canopy and soil surface with the corrected thermal infrared temperature image can accurately extract the maize field T_c and T_soil(T_c:R²≥0.82,RMSE≤0.86℃;T_soil:R²≥0.83,RMSE≤1.07℃).The T_s,T_c,and T_soil gradually decrease with the growth of maize and are greatly affected by rainfall and irrigation,under the same conditions,T_soil is the largest,followed by surface temperature（T_s）,and Tc is the smallest.（4）Field corn evapotranspiration estimation based on surface energy balance method.To improve the feasibility and accuracy of surface energy balance method to estimate evapotranspiration,based on extracting h,LAI,T_s,T_c,and T_soil from the maize field by UAV remote sensing technology,the single-layer energy balance method,and dual-layer energy balance method were respectively improved,and the evapotranspiration estimated by the improved single-layer energy balance method and dual-layer energy balance method based on UAV remote sensing is similar to the calculation results of the original method（single-layer energy balance method:R²≥0.96,RMSE≤0.53 mm;dual-layer energy balance method:R²≥0.95,RMSE≤0.36 mm）.It realizes the rapid and accurate estimation of field corn evapotranspiration based on the energy balance method by UAV remote sensing,especially the improved dual-layer energy balance method,which simplifies the complexity of the dual-layer energy balance method to estimate the field evapotranspiration.（5）Validation of remote sensing estimation method for evapotranspiration unmanned aerial vehicles and analysis of influencing factors.Firstly,based on the water balance method,the different maize field evapotranspiration estimation methods improved based on the drone remote sensing technology that was verified.Then,the effects of irrigation and rainfall on evapotranspiration in a maize field and the effects of different crops and spectral parameters on evapotranspiration in a maize field at different growth stages were analyzed.The results show that the improved dual crop coefficient method can estimate the evapotranspiration of field maize accurately（2018:R²≥0.92,RMSE≤0.88 mm;2019:R²≥0.91,RMSE≤0.50 mm）.Based on a drone remote sensing technology,the spatial distribution map of field evapotranspiration at different growth stages of maize is generated,which can reflect the spatial difference of field evapotranspiration at different growth stages of maize.From the rapid growth period of maize to the late growth period,the sensitivity of evapotranspiration to the effects of rainfall and irrigation first increases and then decreases,and reaches the maximum during the transition period from the rapid growth period to the middle growth period.During irrigation and rainfall,field evapotranspiration began to increase and reached a peak on the 2nd or 3rd day.Besides,it is determined that the dominant factors affecting the field evapotranspiration in different crop parameters and vegetation indices during the rapid growth period are h,and the middle and late growth periods is EVI.