Study On The Classification And Spatialand Temporal Changs Of Main Tree Species In The Greater Xing'an Mountains Based On Landsat Data

The survival and development of any organism depends on the forest,which is the main vegetation type of terrestrial ecosystem.The internal structure of the forest is complex and diverse,and the ecological change process is very obvious.For global climate change,forests play a crucial role in regulation.When studying the spatial and temporal changes of vegetation,forests have important scientific research significance.The first task is to investigate and real-time dynamic monitoring of forest resources.As China's largest state-owned natural forest area and an important forest resource reserve base,the Greater Xing'an Mountains area is in an important position in China's forestry production.The investigation of forest resource information in the traditional sense cannot ensure the accuracy and real-time of the information obtained.In recent years,the use of remote sensing technology to classify forest types has gained a lot of successful experience,which is conducive to the development of forest information update in China and has accelerated the construction of China's forestry ecological environment system and digital forestry.Therefore,the use of remote sensing technology to classify and dynamically monitor the main types of forest trees in the Greater Xing'an Mountains region plays an important role in monitoring and regulating forest resources in China and regulating sustainable forest development.In this paper,Landsat remote sensing image is used as the main data source,and other auxiliary data such as DEM are used,and the Daxinganling area of Heilongjiang Province is used as the research area.Data preprocessing is performed on the remote sensing image;CART decision tree classification method,maximum likelihood method and decision tree The classification method is to classify and extract the typical feature information in the study area,use the measured data to verify the accuracy of the classification results,and perform post-processing such as mosaic mosaic to obtain the classification results of the study area.The main research results include:(1)In the classification of tree species in the Greater Xing'an Mountains area,the study is divided into 7 categories.The decision tree classification method based on vegetation index is used for classification.The overall classification accuracy is75.46%,which is the lowest accuracy.The maximum likelihood method has an overallclassification accuracy of 82.9%,but there are also many wrong and missing points in the results;The machine learning CART decision tree algorithm is used to automatically classify the classification rules according to the sample data.The overall classification accuracy is 86.62%,the Kappa coefficient is 0.84,the highest classification accuracy,and the Pinus sylvestris var.Mongolica and Betula platyphylla are particularly well extracted.In the classification of tree species,CART decision tree classification has a higher advantage than traditional classification methods.(2)The CART decision tree classification method with the best classification accuracy is used to classify the features in the Greater Xing'an Mountains in2005,2010,and 2015.On this basis,the forest vegetation types at different latitudes and different underlying surfaces during 2005-2015 were studied Changes in the area.The total forest area increased by 5353.92km2,the area of larch and birch increased,and the area of Pinus sylvestris was decreasing.Pinus sylvestris var.Mongolica increased in the latitudes of N54°00? ~ 53°00?,and the areas of other latitudes were decreasing.Pinus sylvestris was moving towards higher latitudes;larch was in N53°00?~52°00? and The area of the latitude zone of N52°00?~51°00? increases,the area of the latitude zones of N54°00?~53°00? and N51°00?~50°00? decreases,and the national forest protection plan has achieved good results;the area of birch is in each latitude zone Are all growing.(3)Combining the 2005,2010 and 2015 Daxinganling forest tree classification maps and the Daxinganling area different frozen soil type classification maps.By analyzing the spatiotemporal changes of forest tree species in different permafrost regions in the Greater Xing'an Mountains from 2005 to 2015,it was found that the island-shaped permafrost areas of birch and larch showed an increasing trend in the three-stage classification results,while the area of Pinus sylvestris showed a downward trend;In the three-phase results,the area of birch and Pinus sylvestris in the large continuous permafrost area showed an increasing trend,while the area of larch decreased;in 2005,the large continuous permafrost area of larch was larger than that of the island-shaped permafrost area.Larch occupies more area,but in 2015 the island-shaped permafrost area larch occupies more area than the large continuous permafrost area;and for Pinus sylvestris var.In both,birch is an island-shaped permafrost area that grows more than a large continuous permafrost area.