Estimating aboveground biomass for different forest types based on Landsat TM measurements

Abstract

Forest aboveground biomass (AGB) is an important variable for evaluating ecosystem functions, assessing fire behaviors and impacts, and understanding global carbon balance. Remote sensing technology provides a feasible way to acquire forest stand information at a reasonable cost with acceptable accuracy. This study utilized reflectance in six non-thermal Landsat TM bands and a variety of vegetation indices to identify the relationships between TM data and AGB for different forest types. The field AGB data for testing and validation was from Forest Inventory and Analysis (FIA) datasets of Georgia forests. The forests were classified to softwoods, hardwoods and mixed forests. The strength of correlation between AGB and TM reflectance and vegetation indices was calculated. Multiple regression analyses were used to develop AGB estimation models. The results indicated that vegetation index was better predictive variable than TM single band reflectance in AGB estimation. The vegetation indices including three or more TM bands were more strongly correlated with AGB and more commonly used in AGB estimation models. Different forest types have different relationships between TM data and AGB. The best TM bands in AGB estimation for different forest types are: TM7 and TM1 for hardwoods forests, TM1 and TM5 for softwoods forests, TM3 and TM5 for mixed forests. Potential errors in our AGB estimates could be associated with effects of soil background, the accuracy of land cover data and sampling errors. The possible way to improve the estimation accuracy can be integration of different sources of remotely sensed data or more stand structure information.