Multi-scaled modeling of carbon cycle components in a temperate, deciduous forest

Three studies designed to provide information on important components of regional, deciduous forest carbon cycling are described. These studies utilize ground-based surveys and remote sensing to develop estimates of: landscape-scale carbon stocks, canopy scale Leaf Area Index (LAI) and regional scale aboveground biomass in south-central Indiana. The total carbon content for the study stands ranged from 10.91 ± 0.84 kg m−2 in a 13-year old site to 21.82 ± 5.23 kg m−2 in a 102-year old site. For all stand ages, aboveground live vegetation was the largest carbon pool, containing 45.5% of the total carbon. The mineral soil pool contained 35.9% of the total carbon. Roots contained 9.8% of the total carbon; and the forest floor contained the remaining 8.8%.; Handheld LAI measurements were used with a time sequence of Landsat Thematic Mapper imagery to create models of canopy scale LAI. LAI is most reliably estimated using imagery from the full leaf-out period at the peak of the growing season. Topography is a confounding factor, as LAI for stands on north-facing aspects were more correlated with spectral reflectance than was the case for south facing stands. Indices with spectral information from the near-infrared were more useful than those with only visible wavelengths. TM band 7 was correlated with LAI at peak leaf-out. Tassled Cap transformations of the data were also useful indicators of LAI. NDVI measures were not correlated with LAI.; Two JERS-1 SAR images were used with aboveground biomass estimates from the carbon stocks survey to determine whether this data could be used to estimate forest biomass. In all, sixteen regression models in both linear and non-linear forms were developed to determine whether JERS-SAR imagery could be used for forest biomass estimation in our south-central Indiana study area. None of the models produced were statistically significant at p <= 0.05. Therefore, at the present time, available spaceborne SAR data does not have the ability to serve as a means of forest biomass estimation in this region. Reasons for this lack of correlation are described, along with a discussion of future developments that may alleviate these problems.