Abstract: By the method of harvesting sampling trees, the information of aboveground biomass in 34 plots of a Larix olgensis plantation were collected, of which, the information from 29 plots was selected and fitted with stand factors and TM image RS factors, respectively to establish biomass models, and the information from the rest 5 plots was used to verify the models accuracy. The aboveground biomass in the Larix olgensis plantation could be linearly fitted with either stand factors or RS factors. For the young-middle aged trees, the estimation accuracy of stand factors model was higher(P_stand=94.33%), and the test error was smaller (MRE_stand=6%), compared with RS factors model (P_RS=92.32%, MRE_RS=31%). If only the middleaged trees were taken into account, the estimation effect of the two models had no significant different (error sum E^_stand=329.9 t, E_RS=313.6 t). Overall, the stand factors model was better for the estimation of Larix olgensis cortex, wood, and stumpage biomass, while the RS factors model was better for the estimation of middle-aged trees leaf, flower, fruit, branch, and crown biomass.

Key words: Larix olgensis, stand survey factor, TM image RS factor, biomass model, non-point nitrogen loss, spatial differentiation, multiple time scales, SWAT, Shanmei Reservoir watershed.