Forest structure and aboveground biomass in the southwestern United States from MODIS and MISR

Authors

Mark Chopping, Montclair State University
Crystal B. Schaaf, Boston University
Feng Zhao, Boston University
Zhuosen Wang, Boston University
Anne Nolin, Oregon State University
Gretchen G. Moisen, Rocky Mountain Research Station
John V. Martonchik, NASA/Jet Propulsion Laboratory
Michael Bull, NASA/Jet Propulsion Laboratory

Document Type

Article

Date of this Version

2011

Citation

Remote Sensing of Environment 115 (2011) 2943–2953; doi:10.1016/j.rse.2010.08.031

Abstract

Red band bidirectional reflectance factor data from the NASA MODerate resolution Imaging Spectroradiometer (MODIS) acquired over the southwestern United States were interpreted through a simple geometric–optical (GO) canopy reflectance model to provide maps of fractional crown cover (dimensionless), mean canopy height (m), and aboveground woody biomass (Mg ha⁻¹) on a 250 m grid. Model adjustment was performed after dynamic injection of a background contribution predicted via the kernel weights of a bidirectional reflectance distribution function (BRDF) model. Accuracy was assessed with respect to similar maps obtained with data from the NASA Multiangle Imaging Spectroradiometer (MISR) and to contemporaneous US Forest Service (USFS) maps based partly on Forest Inventory and Analysis (FIA) data. MODIS and MISR retrievals of forest fractional cover and mean height both showed compatibility with the USFS maps, with MODIS mean absolute errors (MAE) of 0.09 and 8.4 m respectively, compared with MISR MAE of 0.10 and 2.2 m, respectively. The respective MAE for aboveground woody biomass was ~10 Mg ha−1, the same as that from MISR, although the MODIS retrievals showed a much weaker correlation, noting that these statistics do not represent evaluation with respect to ground survey data. Good height retrieval accuracies with respect to averages from high resolution discrete return lidar data and matches between mean crown aspect ratio and mean crown radius maps and known vegetation type distributions both support the contention that the GO model results are not spurious when adjusted against MISR bidirectional reflectance factor data. These results highlight an alternative to empirical methods for the exploitation of moderate resolution remote sensing data in the mapping of woody plant canopies and assessment of woody biomass loss and recovery from disturbance in the southwestern United States and in parts of the world where similar environmental conditions prevail.