Toward regional- to continental -scale estimates of vegetation canopy height: An empirical approach based on data from the Shuttle Radar Topography Mission.

This dissertation investigates the feasibility of exploiting interferometric synthetic aperture radar (InSAR) data acquired during the 2000 Shuttle Radar Topography Mission (SRTM) for the purpose of obtaining regional- to continental-scale estimates of vegetation canopy height. The specific objectives were to (1) assess the quality of SRTM C- and X-band data in the context of canopy height retrieval with an emphasis on vertical accuracy and horizontal resolution, (2) determine the extent to which SRTM C-band data could be used to develop empirical models for canopy height prediction, and (3) develop a robust SRTM-based approach for generating a year-2000 baseline map of canopy height for the conterminous U.S. The assessment of SRTM data quality revealed the presence of a vegetation signal sufficient to support canopy height retrieval. In the vertical dimension, signal quality was found to be most affected by error attributed to residual phase noise, and a novel strategy for error mitigation was developed. In the horizontal dimension, the resolution of the SRTM C- and X-band data was estimated at approximately 45 meters. Pilot studies conducted in Georgia and California demonstrated that empirical estimates of canopy height could be obtained from the SRTM C-band vegetation signal in conjunction with the National Elevation Dataset assuming the availability of sufficient field reference data and an appropriate level of error mitigation. The studies also revealed the importance of stand-level characteristics, including stand size and shape in the context of phase noise reduction and stand structure where regression model development is concerned. Supported by an unprecedented confluence of national geospatial data layers as well as an extensive national reference data network, a proof-of-concept study was designed to evaluate a novel, empirical approach for broad-scale SRTM-based canopy height mapping. The study produced the first-ever InSAR-based map of canopy height spanning 62,000 km2 in central Utah. Average absolute height errors resulting from the application of a four-variable, tree-based regression model were on the order of 2.1 m (r = 0.88). This research has culminated in the development of a robust strategy for generating a high-resolution, year-2000 baseline estimate of canopy height for the conterminous U.S.PhDBiological SciencesEarth SciencesForestryGeophysicsRemote sensingUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/125917/2/3224777.pd