An efficient regression strategy for extracting forest biomass information from satellite sensor data

Abstract

Monitoring of biomass in forest ecosystems is important as forest is diminishing rapidly in many parts of the world, which is one of the major sources of global carbon emission. Remote sensing is a useful tool for rapid estimation of biomass. Most of the studies currently available to assess biomass from satellite sensor data using regression provide low correlation. The study explores the possibilities to increase it. Various spectral channels and transformations of Landsat Enhanced Thematic Mapper Plus (ETM+) data for predicting biomass in a tropical forest ecosystem of south‐eastern Bangladesh were tested. One of the interesting findings of the study is the incorporation of dummy variables based on forest types can dramatically increase the correlation.