Abstract: Landsat Thematic Mapper (TM) image was used to estimate Moso bamboo forest biomass, and six atmospheric calibration methods (FLAASH model, 6S model, and DOS1-4 models) were adopted to analysis the effects of atmospheric calibration on the remote sensing estimation of Moso bamboo forest biomass. All the six calibration methods could effectively reduce the atmospheric impacts on TM spectral responses. The relationships between NDVI and Moso bamboo forest biomass under the calibration by the six calibration methods were improved. Great differences were observed in the relationships of Moso bamboo forest biomass with NDVI, II , and MI when using the same calibration methods, suggesting that atmospheric calibration should be made for studying the biophysical significance of vegetation indices. The Landsat TM data corrected with DOS3 model had the highest correlation coefficient with Moso bamboo forest biomass, but there were no significant differences in the correlation coefficients after corrected with the six calibration methods, which indicated that atmospheric calibration might be not required if a single TM image was used for biomass estimation with multiple linear regression model.

Key words: atmospheric calibration, biomass, remote sensing estimation, vegetation index, land surface temperature (LST), MODIS, land use type, Tianshan Mountains area.