Estimation of Forest Height and Canopy Density From a Single InSAR Correlation Coefficient

Journal article, 2015

A two-level model (TLM) is introduced and investigated for the estimation of forest height and canopy density from a single ground-corrected InSAR complex correlation coefficient. The TLM models forest as two scattering levels, namely, ground and vegetation, separated by a distance Delta h and with area-weighted backscatter ratio mu. The model is evaluated using eight VV-polarized bistatic-interferometric TanDEM-X image pairs acquired in the summers of 2011, 2012, and 2013 over the managed hemi-boreal test site Remningstorp, which is situated in southern Sweden. Ground phase is removed using a high-resolution digital terrain model. Inverted TLM parameters for thirty-two 0.5-ha plots of four different types (regular plots, sparse plots, seed trees, and clear-cuts) are studied against reference lidar data. It is concluded that the level distance Delta h can be used as an estimate of the 50th percentile forest height estimated from lidar (for regular plots: r > 0.95 and root-mean-square difference (sigma) < 10%, or 1.8 m). Moreover, the uncorrected area fill factor eta(0) = 1/(1 + mu) can be used as an estimate of the vegetation ratio, which is a canopy density estimate defined as the fraction of lidar returns coming from the canopy to all lidar returns (for regular plots: r > 0.59 and sigma approximate to 10%, or 0.07).