Monitoring tree moisture using an estimation algorithm applied to SAR data from BOREAS

Abstract

During several field campaigns in spring and summer of 1994, the NASA/JPL airborne synthetic aperture radar (AIRSAR) collected data over the southern and northern study sites of BOREAS. Among the areas over which radar data were collected was the young jack pine (YJP) tower site in the south, which is generally characterized as having short (2-4 m) but closely spaced trees with a dense crown layer. In this work, the AIRSAR data over this YJP stand from six different dates were used, and the dielectric constant and hence the moisture content of its branch layer components were estimated. The approach was to first derive a parametric scattering model from a numerical discrete-component forest model, which is possible if the predominant scattering mechanism can be identified. Here, a classification algorithm was used for this purpose, concentrating on areas where the volume scattering mechanism from the branch layer dominates. The unknown parameters mere taken to be the real and imaginary parts of the dielectric constant, from which the moisture content can be derived. Once the parametric model was derived, a nonlinear estimation algorithm was employed to retrieve the model parameters from SAR data. This algorithm is iterative, and takes the statistical properties of the data and unknown parameters into account. The inversion process was first verified using synthetic data. It was observed that the algorithm is robust with respect to the a priori estimate. The estimation algorithm was then applied to AIRSAR data of BOREAS. The results show how the environmental conditions affected the moisture state of this forest stand over a period of six months. It is observed that canopy moisture increased during the thaw season, was stable starting from the end of the thaw season throughout most of the growing season, after which a period of dry-down was observed at the end of the growing season.