Estimation of Forest stands Volume and Tree density using ETM+ and ancillary data

摘要

Sustainable management and utilization of forest resources require accurate inform ation about forests stand characteristics. Remote sensing studies of forest stand mapping can benefit greatly from careful application of digital ancillary data . In this study, we investigated the effect of applying primary topography parameters, which are more geospatially related to forest stand i.e. elevation, slope, and aspect to improve estim ation of forest stand volume and density using ETM+ data. Multivariate regression techniques we re used to predict stand volum e and tree density. The results of multivariate regression analysis using only ETM+ data showed (adjusted R~2=43%; RMSE=97.4 m~3/ha) and (adjusted R~2=73A%; RMSE=170.13 n/ha) for stand volum e and tree density, respectively. By adding the DEM, slope, and aspect param eters to multivariate analysis, RMSE and adjusted R~2 of the models were im proved. Compared with spectral data, these spatial parameters could improve the results so that the adjusted R~2 increased from 43% to 57.3% and 73.4% to 79% for stand volume and tree density, respectively. In addition, the RMSE and Bias decreased from 97.4 and 28.08 (m~3/ha) to 81.1 and 13.48(m~3/ha), 170.13 and 61.475 (n/ha) to 168.69 and 49.6 (n/ha) for stand volum e and tree density, respectively. Adding ancillary data that spatially related to forest stands may increase accuracy of estimations.