Extraction of Forest Coverage with Remote Sensing Image Classification

Forest coverage rate is an indicator that reflects the possession of forest area or the abundance of forest resources and the degree of greening in a country or region and is also an important basis for determining the policy of forest management and development and utilization. For example, the extraction of the forest coverage rate is the basic data for the back volume and biomass inversion. Based on this, this paper takes Shangri-La of Yunnan Province as the research area, adopts the fusion of multispectral and panchromatic images, and extracts the forest coverage rate based on the classification technology. The data source includes Landsat 8 data, ZY3-01, GLAS 14, DEM and other auxiliary data, to extract features such as spectrum, NDVI etc., utilizing high-precision GLAS 14 data and SVR regression method to correct the ground objects height obtained by ZY3-01, incorporating this spatial feature into spectral features and then using Classification And Regression Trees (CART) method to classify them. Yearbook statistics are evaluated and verified. The results showed that the forest coverage rate of Shangri-La forest land extracted by the CART method was 74.22%. The forest resource planning and design investigation report in Shangri-La County, Yunnan Province is 74.99%, which shows that the CART decision tree method in this study area is applicable to the extraction of forest coverage rate, and it also proves that the classification methods of remote sensing data is efficient for the ecological indicator inversion.