The Performance of Relative Height Metrics for Estimation of Forest Above-Ground Biomass Using L- and X-Bands TomoSAR Data

Both synthetic aperture radar tomography (TomoSAR) profiles and light detection and ranging (LiDAR) waveforms are the responses of a 3-D canopy structure. Relative height (RH) metrics are extensively applied for forest biophysical parameters [i.e., the forest height and above-ground biomass (AGB)] estimation in full-waveform LiDAR studies. However, the use of RH metrics to forest biophysical estimation with TomoSAR profiles is limited due to the estimation error in the ground peak. To overcome this problem, RH metrics were redefined to avoid directly estimating the ground peak in this article. The redefined RH metrics were examined based on the
L
- and
X
-bands multibaseline (MB) SAR data simulated by LandSAR, which was a coherent backscattering model of 3-D forest canopies with the capability of MB data simulation at a landscape scale. First, the performance of LandSAR in modeling the tomographic features was verified over mountainous areas using the reference DSM and LiDAR digital terrain model acquired in 2012 in the frame of the Daxinganling campaign. Subsequently, the tomographic profiles were retrieved from the simulated MB data by using the Capon method. Additionally, redefined RH metrics were derived and then applied for retrieving the forest height. The highest performance corresponded to the combination of the
L
-band redefined RH metrics, with a correlation of
R
²
= 0.863. The
X
-band redefined RH metrics performed worst due to limited penetration. Finally, the estimated forest height was used for the AGB retrieval with a height-to-biomass allometry. The
R
²
and root-mean-square error of the forest AGB estimated using the combined model were 0.814 and 28.566 t/ha, respectively, compared with the reference forest AGB. All findings demonstrated that the redefined RH metrics had the potential of forest height and AGB retrieval using low frequencies TomoSAR data.