The Forest Carbon Stock Models In Heilongjiang Province

The carbon stock prediction model on stand level was developed based on the relationship between stand variables and carbon stocks and the forest continuous inventory data from2005²010.The weighted nonlinear least square was used to eliminate the heteroscedasticity for the models of the different origins.Due to the difference between regions,the carbon of the same forest types may be different.Therefore,the carbon stock models of different regions were developed by using the dummy variable approach,Thus,the prediction of the carbon reserves of forest in different regions and origins is carried out.The results showed that the forest carbon model for natural forest and plantation which distinguished the origins has good performance and R² is greater than 0.94.Model evaluation statistics of mean error percent?ME%?was within the range of±6.00%and the mean absolute error percent?MAE%?are less than 10%except for the 15.33%for black birch natural forest.Most of the model prediction accuracy?P%?is above 95%.With dummy variable method and considering the different regions of Compatible forest carbon,R² are greater than 0.94 and the mean error ME were smaller.The mean absolute error percent MAE%are less than 8%.The prediction accuracy P%are all above 95%.As for the dummy variable used to develop the general model,the carbon stock of the specific region increased with the increasing of value for the parameter b on the condition of the mean basal area and parameter a and c were fixed;the carbon stock increased with the increasing of parameter c for the different regions when the height of forest stand,other stand variables and parameter a and b were constant.The results of this study provide the basis for estimate different forest types of forest carbon in Heilongjiang province,and it provides a reference for the large-scale assessment of forest carbon reserves.