Study On Winter Wheat Yield Estimation In Qitai County Based On Remote Sensing Technology

Wheat is one of the most important food crops for human beings.Now it is widely planted all over the world,and the planting area of winter wheat is more than 80%of the total wheat planting area.How to quickly and efficiently obtain the winter wheat planting information and growth monitoring is very important for the winter wheat yield estimation and food security.In order to extract crop planting structure quickly,Qitai County,the main grain producing area in Xinjiang,was studied by object-oriented identification method,and crop information extraction and remote sensing estimation methods of winter wheat yield were researched.Based on the landsat8 remote sensing images,the land use status map in 2016 was updated by using the 2019 image,and the land use status map in 2019 was obtained.Using ecognition and ESP2 plug-in to determine the optimal segmentation scale of the study area,combinning with the field survey data,the main crops in Qitai county were divided into four categories:wheat,corn,melon and sunflower by using the object-oriented cart decision tree classifier and random forest classifier,and the crop planting information of Qitai county was extracted.Based on the field measured data and remote sensing data,the yield estimation model,remote sensing yield estimation model and remote sensing actual measurement comprehensive yield estimation model are established.The results show that:(1)The best segmentation scale of remote sensing image in the study area was calculated by ESP2,and the best segmentation scale was determined as 90.In this experiment,the classification accuracy was higher when the number of cart trees in random forest classifier was 80-90;the overall accuracy of cart decision tree was 0.925,kappa coefficient was 0.893;the overall accuracy of random forest classifier was 0.945,kappa coefficient was 0.921.The medium spatial resolution remote sensing image can be used to identify the county level crops quickly and effectively;the accuracy of winter wheat identification in Qitai county in 2019 will reach 98.64%.(2)This study shows that the vegetation index at the heading stage and the measured factors during the filling stage have the highest correlation with the yield.Therefore,the heading to the filling stage is the best period for modeling winter wheat yield.(3)Among the 10 remote sensing vegetation indexes,NDVI,GNDVI and RVI vegetation indexes have the highest correlation with yield;among the remote sensing yield estimation models,actual yield estimation models,and remote sensing actual measurement comprehensive yield estimation models,the accuracy of remote sensing actual measurement comprehensive yield estimation model is 90.93%,and the applicability of remote sensing yield estimation model is the best.