Research On Forest Resource Changes Monitoring Based On Multi-source Remote Sensing Images

As the largest ecosystem on the planet,forests play a vital role in purifying the air,regulating the climate,preventing soil and water loss,and maintaining biological species diversity.The state has accordingly launched projects like the Grain for Green Project to protect forest resources,the Loess Plateau which has a fragile ecological environment has become one of the key areas under the the Grain for Green Project.In order to accurately monitor the distribution of forest resources and analyze the dynamic changes of forest resources in the Loess Plateau,the PLASAR radar remote sensing data with a spatial resolution of 25 meters and Landsat optical remote sensing data with a spatial resolution of 30meters were selected.The data fusion algorithm of the two data sources was designed to extract annual forest resource distribution information in the Loess Plateau;The accuracy of the PL-based(PALSAR/Landsat)forest product maps obtained from the fusion of the Landsat and PALSAR was compared with that of eight mainstream medium-high-resolution forest product maps(China Cover,FROM-GLC,GLCF,Globe Land30,Hansen,JAXA,NLCD,OU-FDL)on the market,the calculation of consistency between forest product maps on a pixel scale and the reasons for the uncertainty between forest product maps,the PL-based forest product map was verified for changes;Finally,the PL-based forest product map was used to analyze the spatiotemporal changes of forest resources in the Loess Plateau and the Wuqi county which is a a key county in the Loess Plateau.The results show that:(1)The PL-based forest product map using the forest structure information from the vertical direction of the PALSAR radar remote sensing data and the NDVI spectral information from the horizontal direction of the Landsat optical remote sensing data had the highest overall accuracy(91.27%)and Kappa coefficient(0.7728).(2)The overall consistency of the PL-based forest product map with FROM-GLC forest product map based on single optical remote sensing data Landsat,OU-FDL forest product map integrating radar remote sensing data PALSAR and optical remote sensing data MODIS,and JAXA forest product map based on single radar remote sensing data PALSAR was 91.35%,93.59%and92.23%respectively on the pixel scale,in which the forest area accounted for 11.47%,9.85%and 7.95%,and the non-forest area accounted for 88.53%,90.15%and 92.05%.(3)Analysis of the spatial and temporal changes of forest resources in the Loess Plateau through the PL-based forest product map,the forest area increased from 1.05×10~5 square kilometers in 2007 to 1.20×10~5 square kilometers in 2017,and the annual average forest area increased by 1500 square kilometers.(4)The PL-based forest product map was used to analyze the achievements of Grain for Green Project in Wuqi county,the forest area increased 353.75 square kilometers from 2007 to 2017 and the the afforestation effect was remarkable.