Remote Sensing Retrieval Of Canopy Water Content For Winter Wheat In Growing Period Based On Evolutionary Algorithm

Canopy water content is an important indicator of crop health,timely and accurate acquisition of canopy water content in a wide range of winter wheat growth period is conducive to real-time acquisition of wheat growth information,drought stress assessment,and guidance of precision irrigation.Thus,it is important to build high-standard farmland and ensure food security.Traditional field measurement methods are time-consuming and labor-intensive,which are easy to destroy field crops and can only be sampled at a single point.Medium and high resolution remote sensing satellites,Landsat-8 and Sentinel-2,have the advantages of multi spectrum and high timeliness,which provides data support for rapid acquisition of large-scale winter wheat canopy water content.However,due to the limitation of image spatial resolution,there are a large number of mixed pixels of wheat and soil during the growth of winter wheat,and the soil has a great impact on the retrieval of wheat canopy water content,which is difficult to eliminate.In addition,there are few retrieval studies on winter wheat in the whole growth period.Based on this,this research takes Wuji county and surrounding farmland in Hebei Province as the research area,integrates the complementary advantages of Landsat-8 and Sentinel-2data,and carries out the inversion of winter wheat canopy water content in the whole growth period.The following work has been carried out:（1）Extraction of winter wheat planting information.The Sentinel-2 image（April18）is classified based on support vector machine to extract the spatial distribution information of wheat field,and the accuracy is 94.5%.The field scale coverage data of wheat were obtained by photography,which was used as the verification data.（2）Based on the vegetation index（NDVI,MSAVI,EVI）,the vegetation coverage in different growth stages of winter wheat was retrieved by regression model and pixel dichotomy model.Through the verification with the measured data,the results obtained by the pixel dichotomy model based on EVI index have high accuracy,which can reflect the changes of winter wheat vegetation coverage in different growth periods and can be used as the key parameter to quantify the soil effect for subsequent inversion.（3）Based on the idea of mixed pixel decomposition and taking vegetation coverage as a quantitative index,a mixed pixel linear decomposition model is established.The field water content is input as the measured water content of wheat field,establishing the regression model between the simulated value and four vegetation water indexes（RMSI,SRWI,NDWI,NDII）,and the optimal regression model is then selected to construct the retrieval equation,which is optimized and solved by evolutionary algorithm.The comparison between the retrieval results and the measured results shows that the retrieval results of Sentinel-2 data have the highest accuracy.Among them,the retrieval result of particle swarm optimization algorithm（R~2=0.762,RSME=4.0%）is better than those of genetic algorithm（R~2=0.764,RSME=4.7%）.The results show that quantifying the linear mixing ratio of wheat canopy and soil background can effectively eliminate the influence of soil on the retrieval of wheat canopy water content and improve the retrieval accuracy.This paper proposed a retrieval scheme based on the mixed pixel linear decomposition model and evolutionary algorithm,which can quickly and accurately retrieve the canopy water content of large-area winter wheat.The multi temporal retrieval of winter wheat canopy water content in different growth periods in North China is realized,which provides technical support for real-time monitoring of wheat growth period.It further shows that using medium and high-resolution optical remote sensing data（Landsat-8 and Sentinel-2）to monitor crop canopy water content has important application potential.