Date of Award
Program or Major
Doctor of Philosophy
Foliar nitrogen concentration represents a direct and primary link between carbon and nitrogen cycling in terrestrial ecosystems. Although foliar N is used by many ecosystem models to predict leaf-level photosynthetic rates, it has rarely been examined as a direct scalar to stand-level carbon gain. Significant improvements in remote sensing detector technology in the list decade now allow for improved landscape-level estimation of the biochemical attributes of forest ecosystems.
In this study, relationships among forest growth (aboveground net primary productivity (ANPP) and aboveground woody biomass production (AWBP)), canopy chemistry and structure, and high resolution imaging spectrometry were examined for 88 long-term forest growth inventory plots maintained by the USDA Forest Service within the 300,000 ha White Mountain National Forest, New Hampshire.
Analysis of plot-level data demonstrates a highly predictive relationship between whole canopy nitrogen concentration (g/100 g) and aboveground forest productivity (ANPP: R2 = 0.81, p < 0.000; AWBP: R 2 = 0.86, p < 0.000) within and among forest types. Forest productivity was more strongly related to mass-based foliar nitrogen concentration than with either total canopy N or canopy leaf area.
Empirical relationships were developed among spectral data from the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) and field-measured canopy nitrogen concentration (mass basis). Results of this analysis suggest that hyperspectral remote sensing can be used to accurately predict foliar nitrogen concentration, by mean of a full-spectrum partial least squares calibration method, both within a single scene (R2 = 0.84, SECV = 0.23) and across a large number of contiguous images (R2 = 82, SECV = 0.25), as well as between image dates (R2 = 0.69, SECV = 0.25).
Forest productivity coverages for the White Mountain National Forest were developed by estimating whole canopy foliar N concentration from AVIRIS spectral response. Image spatial patterns broadly reflect the distribution of functional types, while fine scale spatial variation results from a variety of natural and anthropogenic factors. This approach provides the potential to increase the accuracy of forest growth and carbon gain estimates at the landscape level by providing information at the fine spatial scale over which environmental characteristics and human land use vary.
Smith, Marie-Louise, "Landscape-scale prediction of forest productivity by hyperspectral remote sensing of canopy nitrogen" (2000). Doctoral Dissertations. 2129.