Estimating Forest Biomass Using MODIS and FIA Data in the Lake States: MN, WI, and MI, USA

This study linked the Moderate Resolution Imaging Spectrometer (MODIS) and USDA Forest Service, Forest Inventory and Analysis (FIA) data through empirical models established using high-resolution Landsat ETM+ observations to estimate aboveground biomass (AGB) in three "Lake" States in the north central USA. While means obtained from larger sample sizes in FIA datasets can be used as reference numbers over large scales, remote sensing (RS) based observations have the ability to reflect spatial variation of interested properties within a given area. Thus, combining two national on-going datasets may improve our ability to accurately estimate ecological properties across large scales. Using standard and consistent data sources can reduce uncertainty and provide more comparable results at both temporal and spatial dimensions. We estimated total forest AGB in the region was 2,548 million tons (dry weight) in 2001 with mean AGB value of 95 Mg/ha ranging from 4 to 411 Mg/ha (within 95% percentiles). Mixed forests featured 66% of the total AGB while deciduous and evergreen forests contained 32% and 2% of the total AGB, respectively, at 1-km pixel resolution. Spatially, AGB values increased from northwest to southeast in general. The RS-based estimates indicated a greater range in AGB variations than the FIA data. Deciduous forests were more variable (both in absolute and relative terms) than evergreen forests. The standard deviation of AGB for deciduous forests was 137 Mg/ha, or a coefficient of variation of 92%; while that for evergreen forests was 24 Mg/ha, or a coefficient of variation of 37%.