Exploring the effect of forest spatial heterogeneity using coherent three-dimensional radar backscattering model

Modeling natural forest is challenging since the vertical and horizontal structural heterogeneity of variations of tree height, scatter shape, and pattern resulted from different tree species, growth stage, and environmental variables. The heterogeneous effect of mix-scene is not well understood quantitative so far, which might cause substantial uncertainty in the soil moisture and biomass estimations. We investigated effects of trunk tilt, trunk tapering, existence of understory and mix-scenes of forest combined with grass by using a coherence 3D vegetation radar model, which is firstly parametrized and validated using the ground measurements and airborne P band SAR data at pine forest of Metolius, Oregon, USA, 2013-2014. The airborne SAR data and simulated data are in good agreement with mean difference less than 1dB and correlation coefficient above 0.96. The tilt angle from 0 to 10 degree can reduce the backscatter coefficient of 1.55dB and 1.03dB for HH and VV, respectively. Tapering of trunk causes the decrease of 2.54dB and 6.35dB for HH and VV polarizations. Simulations of several mix-scenes of the same tree and ground parameters but difference tree clumping patterns indicate the largest difference can up to 3dB of extreme tree clumping patterns.