Study On Estimation Of Forest Volume Based On RS And GIS

Forests are important resources for national construction, while the quantitative estimation of forest volume has been the focus of the study of forestry scientists. The traditional methods of estimating forest volume is a kind of work which has a long cycle, needs heavy tasks and intensive labor and will need the massive funds. How to use the RS and GIS technology, combined with a small amount of ground-to information, to establish the estimated equation of the forest stock volume in the monitoring regions to estimate the forest stock volume, can minimize the work load of ground field investigation. Nowadays, it has become a hot issue on forestry.In this paper, the study area is Laoshan Forest Farm in Jiangsu Province. This study mainly used RS and GIS technology and took the TM images and forest resources survey data as the basic data source, to explore the process of remote sensing estimation of the forest volume and to establish an estimation model. The main content s and results of this research are summarized as follows:(1)First, this study made geometric correction and atmospheric correction on the TM images. Through the analysis of the statistical characteristics of band and spectral characteristics of forest types, the results show that the TM4 band has the largest amount of information and the physical characteristics are more obvious.(2)Collect the field survey data, set the whole forest farm as overall, and small class as a unit. Set and extract the small class of RS and GIS factor. The TM factor is the band reflectance values and the reflectance ratio, GIS factor is the terrain extraction of high-precision DEM.(3)Set RS and GIS factors as independent variables,the average hectare volume as dependent variable to build the multi-regression mode based on estimates of the least-squares. Sample data were selected by standard deviation law and factor variables were selected by the method of principal components analysis, stepwise and enter regression analysis. The average precision of the model is 88.20%。...