Remote Sensing Estimation Of Forest Carbon Storage In Baima Forest Farm

Taking Baima Forestry Station as study area and using2013Landsat8OLI remote sensing image as data source combined with field survey data, the study established the inversion model of forest biomass and carbon storage based on remote sensing and GIS technique, in order to estimate forest biomass and carbon storage of Baima precisely effectively and offer practical experience in similar research in China.After a series of pre-processing for image like geometric correction, image projection transformation, radiation calibration, image fusion, using c terrain correction method to eliminate influence of terrain on radiation, and FLASH atmospheric correction method converts radiance to the real surface reflectance. Vegetation index as NDVI, DVI etc. were calculated. Supervised classification was carried out for surface features, and the classification results were corrected utilizing of artificial interpretation to extract distribution of forest resource. The way of biomass conversion factor was applied to calculate forest biomass of plots according to stand volume from field sample. Then,24independent variables were extracted with the help of ArcGIS Desktop software, including terrain factor such as altitude, gradient, aspect, each band reflectance of remote sensing data, the principal component values, band ratio and vegetation indices.Correlation analysis was carried out between biomass and independent variables to generate the correlation coefficients between each factor and the biomass. According to them, independent variables were chosen for modeling respectively, applying monadic regression analysis, multiple regression analysis, and principal component analysis to establish biomass model. Results are below:(1) Compared with the advantages and disadvantages of each model, it was found that multiple stepwise regression model was more accurate than single variable model, the model satisfied with the demand of F test and D-W test and it had the highest coefficient of determination and performance best. The relative accuracy of estimated biomass through test of the independent plots data is88.2%.(2) Supervision classification result showed that forest area was29836.7hectare. According to the calculation of the best model, there was an estimation of total forest carbon reserves of778100tons, the average of the forest carbon density of26.08tons per hectare. Compared with forest investigation data, the relative accuracy of estimated forest area reached87.4%, while carbon reached82.1%.(3) On the basis of horizontal level, forest carbon reserves are mainly distributed in east, south slope. Carbon density is given priority to with moderate levels, high and low carbon density of forest area is less.(4) On the vertical level, forest carbon was mainly distributed in the600-800m medium altitude. Forest carbon reserves also vary in different slope, forest carbon in gentle and flat slope is higher.