Spatial And Temporal Distribution Characteristics And Yield Analysis Of Oilseed Rape In Hubei Province Based On Mixed Pixels

Oilseed rape is one of the most important oil crops in China,and it is important to master the spatial and temporal distribution information of oilseed rape for the macro regulation of oilseed rape production.With the continuous development of modern technology,the extraction of feature distribution information by remote sensing means instead of traditional manual methods has been commonly applied.MODIS data are widely used in the extraction of spatial distribution information of crop planting because of the characteristics of free policy,high temporal resolution,wide sweep,hyperspectral and generating many products,but the coarse spatial resolution severely restricts its extraction accuracy and causes the problem of mixed image pixels.To solve this problem,this study takes oilseed rape in Hubei Province as the research object,and uses high-resolution Landsat7,ALOS,High-Spectrum-1,Sentinel 2B data and medium-resolution MODIS data within the last two decades（2002,2010,2015,2020）as data sources,and applies Arc GIS,ENVI and IDL language By comparing the classification accuracy of oilseed rape in pure and mixed image elements with different yellow indices,we can determine the feature amount that can accurately reflect the feature of oilseed rape,develop the classification rules to identify oilseed rape in combination with NDVI,obtain the spatial and temporal distribution characteristics of oilseed rape in Hubei Province,and perform accuracy statistics on the classification results to achieve high precision extraction of spatial distribution of oilseed rape cultivation.The results were verified by using previous years’yearbook data,and yield analysis of oilseed rape in Hubei Province was conducted.The main findings of this paper are as follows.（1）A comparative study on the accuracy of remote sensing feature extraction of different yellow indices for oilseed rape classification was completed（The four yellow indices compared are the yellow index calculated by the CIE color space model,the yellow index calculated by the Lab color space model,the normalized difference yellow index,and the ratio yellow index）.The results showed that:the highest KAPPA coefficient（0.75）of YILAB extraction accuracy was applied in the pure image element,indicating the best accuracy of yellow index effect in Lab color space model;in the mixed image element,the abundance in MODIS mixed image element showed a significant linear relationship with the four yellow indices in the mixed image element at the flowering stage of rape,among which YILAB results were much higher than the other three yellow indices(The correlation coefficient was 0.94,which indicated that YILAB could respond well to the abundance of oilseed rape in mixed image elements in Hubei Province.Therefore,whether using pure image elements or mixed image elements for oilseed rape classification,it is recommended to characterize the spatial distribution of oilseed rape by the yellow indices in Lab color space model.（2）The extraction of spatial distribution information of oilseed rape in Hubei Province was completed.The results showed that the oilseed rape in Hubei Province was mainly distributed in the Jianghan Plain,the central-northern region of Hubei and the eastern region of Hubei,and the oilseed rape cultivation area in Hubei Province was gradually reduced from 2010 onward.When the abundance of oilseed rape like elements was greater than 20%,the accuracy of the extracted oilseed rape distribution in Hubei Province was higher than the real distribution.After secondary classification according to different abundance and decomposition of mixed image elements,the classification accuracy of oilseed rape will be further improved.（3）The calculation of oilseed rape acreage and yield analysis in Hubei province were completed.The results showed that in predicting oilseed rape yield,the correlation between YI_Lab and oilseed rape yield in Hubei province was higher compared with the commonly used NDVI values,and the correlation between YI_Lab and oilseed rape yield at flowering stage was higher,so it is suggested that using YI_Labat flowering stage（late March）to predict oilseed rape yield in Hubei province will be more effective.