Forest Classification by Multiple-Forward-Mode 5-Scale Modeling

摘要

The Multiple-Forward-Mode 5-Scale approach (MFM-5-Scale) has been demonstrated as a new approach for unsupervised cluster labeling and landcover classification. The basis for this is its ability to link powerful models of forest stand structure and radiative transfer with satellite image response patterns to provide a flexible, physically-based, semi-automated classification and biophysical parameter estimation algorithm. In this work, results using MFM-5-Scale are compared with those obtained by the Enhancement Classification Method (ECM), a highly accurate yet subjective and labour intensive approach which involves considerable user judgement and expertise. The goal was to approach ECM accuracy using MFM-5-Scale, but without the subjectivity of ECM. A mosaic of 7 Landsat TM scenes from different years and covering most of the BOREAS region in western Canada was used in this study. Classifications of 12 forest classes including deciduous and species-specific coniferous classes subdivided into low, medium and high crown densities as well as a mixed forest class were evaluated against 136 field checked validation sites. Overall classification accuracies were 91% for ECM and 85% for MFM-5-Scale. Additional assessments of agreement between the MFM-5-Scale and ECM products were also performed over larger sample areas. The two products were in 76% agreement for all 12 classes (n=6000), and were 94% in agreement for a smaller sample of 6 forest density classes (n=3730). These results are viewed as quite positive towards defining an operational approach, given the objective, semi-automated nature of MFM-5-Scale compared to subjective, user-driven methods such as ECM.