Modeling and inversion of the radar response of vegetation canopies.

Modern airborne and spaceborne remote sensing radars provide an enormous amount of information about electromagnetic scattering from the earth's surface. Since a large fraction of terrain is covered with vegetation having many different canopy structures and electrical properties, it is important to understand the mechanisms by which such systems scatter radiation. This should permit more effective utilization of the data acquired by such radars. The most popular method currently in use for modeling the radar response of vegetation is radiative transfer (RT). This method has the advantage of being easy to formulate and apply, which makes it convenient for use in inversion algorithms, but the fundamental underlying assumptions on which it is based also limit its applicability in the modeling of vegetation. Five chapters of this thesis are concerned with the vegetation modeling problem. Monte Carlo simulations that account for the second-order coupling between constituents are constructed for trunk dominated vegetation canopies. These simulations provide a benchmark for use in evaluating the performance of RT theory in a regime for which no other analytical models are satisfactory. The interparticle coupling models and simulations are validated using both experimental and numerical methods. A hybrid Monte Carlo/RT model for heterogeneous canopies that can be applied in cases where pure RT theory fails is developed. It is demonstrated in this work that RT theory does not properly predict the scattering behavior of canopies which have high number densities of large scatterers. It is also shown that in such cases multiple scattering and non-uniform illumination invalidate the RT concept of phase and source functions. The seventh chapter of this thesis is devoted entirely to the inverse scattering problem. An iterative inversion algorithm is formulated that may be used with any scattering model. The algorithm is applied to invert the response of the first-order RT model for a general, representative vegetation canopy. A propagation of errors analysis is provided to show the predicted response of the algorithm to errors originating from various sources.Ph.D.Applied SciencesEarth SciencesElectrical engineeringGeographic information science and geodesyUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/129223/2/9409785.pd