Method And Application Of A Multi-parameter Remote Sensing Yield Estimation For Winter Wheat

As a country with a large population and agriculture,ensuring food security is a fundamental national strategy.At present,agricultural information in statistical departments is mainly obtained through catalogue survey methods,and the statistical data reporting process is tedious and time-consuming,and it is difficult to obtain information on the spatial distribution of crops.In this paper,we study the yield estimation of winter wheat using remote sensing technology from the perspective of the business needs of agricultural insurance verification in Henan Province,and construct a winter wheat information extraction model and a yield estimation model,aiming to provide a reference basis for the yield estimation of winter wheat in agricultural insurance verification.The main research contents are as follows:(1)Time-series EVI data reconstructionThe Savizky-Golay filter and the Fourier harmonic time-series filter were used to reconstruct the EVI of the whole growth cycle of winter wheat,and it was concluded through comparative analysis that the Savizky-Golay filter could maximally eliminate noise and retain the original image data.(2)Two winter wheat identification models based on decision trees and deep learning were constructedThe MODIS image and Sentinel-2 image were used as data sources to give full play to the advantages of both and to complement each other.The two classification methods based on decision trees and deep learning were used to construct winter wheat recognition models with different spatial resolutions,and the accuracy of the classification results was evaluated to conclude that the accuracy of the two recognition models for winter wheat can meet the needs of extracting yield parameters in a large range.(3)Construction of winter wheat yield estimation model in Henan ProvinceIn this paper,four parameters,namely enhanced vegetation index(EVI),surface temperature(LST),crop water stress index(CWSI)and trend yield(Yt),were proposed as modeling drivers for the yield estimation model,and a winter wheat yield estimation model was constructed for Henan Province using data from 2013 to 2019,and winter wheat yield in Henan Province in 2020 was estimated.The results show that the accuracy of the estimation model taking into account multiple parameters is more than15% higher than that of the estimation model using only a single parameter,which proves the effectiveness of the method in this paper,and the research results can provide data support for remote sensing verification of agricultural insurance.There are 16 figures,14 tables and 69 references in this thesis.