| As the theme of the ecosystem,the forest is the most complex and widely distributed ecosystem on land.Accurately,timely and efficiently forest mapping has always caught much attention.With the development of remote sensing technology,people can accurately and quickly obtain image data of target objects.Among many developed global land use data,the highest resolution is the Sentinel-2 land use data(10m resolution)released by ESA in 2020.However,the accuracy of the data extraction of ground features needs to be improved.The Yellow River Basin is an important ecological barrier and economic zone,and it is one of the most important human-land coupled regions.In 2021,the"14th Five-Year Plan for Henan Province’s National Land Space Ecological Restoration and Forest Henan Construction Plan"clarified a series of goals such as building the Yellow River Beach area,and determined the overall pattern of forestry protection,development and ecological restoration.However,at present,most scholars still focus on the vegetation in the Yellow River Beach area.In addition,the natural and geographical environment of the Yellow River Beach area is complex,resulting in less research on the Yellow River Beach area.Therefore,it is crucial to obtain the forest information of the Yellow River Beach area accurately,timely and efficiently.In this study,the scope of the Yellow River Beach area is determined based on the Yellow River Levee.Considering that there are still a large number of breakwater forests outside the beach area,the beach area is buffered outward by 500m to carry out the identification and analysis of forest within the research scope.This study converted the RGB color space of 709 sub-meter Google images into HLS and HSV color space,and constructed the Thresholds for Forest and Shadow Indices(FSIT)and the Water and Vegetation Index(WVI)to extract forest,and extract water and farmland data based on Sentinel-2 images,and then remove the confusing water body and farmland in the forest.Mainly completed the following research:(1)According to the color space,construct the FSIT and WVI indices,and completed the preliminary extraction of forest in the Yellow River floodplain.Comparing the extraction results,it is found that the two indices can better identify the woodland and eliminate the interference between the built-up area and the grassland.However,the extraction effect of forest land will be disturbed when there are green water bodies and farmland,resulting in the confusion of forest land,water body and farmland.By comparing the extraction effects of the two indices,this study finally selected 322 images extracted based on the FSIT index and 387images extracted based on the WVI index.(2)Based on the multi-spectral features of sentinel-2 image,this study extracts and eliminates the water information and farmland information within the study area.Finally,the percentage grid product of forest in the beach area of the Yellow River is obtained.(3)Using the methods of confusion matrix and area comparison,it is found that the accuracy of forest extracted by FSIT and WVI index exceeds the accuracy of forest land use data released by ESA.At the same time,it is found that the forest area extracted in this study has a higher correlation with the area proportion of artificial interpretation.For each type of object,200 verification points were selected,and after comparing the high-scoring images,it was concluded that the accuracy of the water body extracted in this study was90%,the accuracy of the water body published by ESA was 89%,and the accuracy of the farmland extracted in this study was 87%.The accuracy of farmland released by ESA was 88%,the accuracy of forest land extracted in this study is 87.5%,and the accuracy of forest land released by ESA was 84%;Combined,a total of 600 verification points were obtained.The accuracy of the water body extracted in this study was 88.3%,the accuracy of the water body released by ESA was 87.83%,the accuracy of the farmland extracted in this study was 83.83%,and the accuracy of the farmland extracted by ESA was 83.5%.the accuracy of the forest extracted in this study is 84.5%,and the accuracy of the forest extracted in this study is 84.35%.After comparison,indicating that the FSIT and WVI indexes can meet the needs of this study and identify the forest in the floodplain with complex environment.(4)According to the extracted forest area data,the total forest area of the Yellow River Beach area is398.86 km~2,accounting for 11.30%of the total beach area.The forest area in the north of the Yellow River accounts for 65.29%of the total area,and the forest area in the south accounts for 34.71%of the total area.The Puyang section of the Yellow River Beach area is the prefecture with the largest forest area,accounting for 29.61%of the total forest area;the Xinxiang section of the Yellow River Beach area accounts for 24.87%of the total forest area;the Zhengzhou section accounts for 17.43%of the total forest area.Among them,Taiqian County is the county with the largest forest area in Puyang City,and it is also the county with the largest forest area among the 19 counties along the Yellow River Beach area,accounting for 54.34 km~2,accounting for 46%of the total forest area of Puyang City.of 13.62%.Based on sub-meter-level high-score images,this paper focuses on the complex natural and geographical environment of the Yellow River Beach area,and constructs a woodland index to identify woodland,which can realize comprehensive monitoring of the forestland in the Yellow River Beach area,strengthen the protection of forestland,and contribute to the protection of China’s ecological security,the biodiversity of the Yellow River Basin,and the ecosystem of the Yellow River floodplain have important practical significance. |
