Inference of forest tree volume using synthetic aperture in central Sudan

The estimation of forest volume plays a crucial role in sustainable management practices aimed at reducing emissions resulting from deforestation and forest degradation. This particular study focused on investigating the correlation between the volume of A. seyal trees and RADAR backscatter, and subsequently deriving tree volume from the backscatter data obtained from RADAR. Field data was collected from two different locations, where systematic samples were established and measurements of tree height (Ht) and diameter at breast height (DBH) were taken. Additionally, Sentinel 1 C-band RADAR (VV, VH polarization), PALSAR, and ALOS 2 L-band (HH, HV polarization) backscatter data were acquired and analyzed to assess their sensitivity in estimating tree biophysical parameters. Land cover maps were then generated using Sentinel 1 data, and tree volume at the second site was obtained using a water cloud model based on ALOS 2 data. In the Wad Elbashir forest, the backscatter data from Sentinel 1 and ALOS PALSAR (cross-polarization) during the dry season exhibited a strong correlation with tree volume (R2 = 0.56 and R2 = 0.70, respectively). However, these relationships were found to be insignificant during the wet season and with like-polarization. In contrast to DBH, tree height (Ht) demonstrated a robust relationship (R2 = 0.60) with sigma-naught for ALOS PALSAR. In Okalma, ALOS 2 HH backscatter data showed a relatively strong correlation with tree volume (R2 = 0.54) compared to HV (R2 = 0.49), and lower R2 values were observed between tree volume and Sentinel 1 data when cross and like-polarization were assessed. Furthermore, tree height (Ht) exhibited a strong correlation with sigma-naught for σ° HH, σ° HV, σ° VV, and σ° VH (R2 = 0.73, R2 = 0.72, R2 = 0.62, and R2 = 0.66, respectively). Furthermore, the utilization of a water cloud model incorporating gaps (with a constant β) fails to accurately estimate tree volume. When β is adjusted based on the backscattering coefficient, it was observed that a linear function overestimated tree volume at higher values, while a quadratic function provided more appropriate estimates. The application of a semi-empirical model known as the Extended Water Cloud Model effectively mapped the volume of the forest. Additionally, it is deduced that RADAR data acquired during the dry season exhibits a correlation with tree biophysical parameters, enabling the retrieval of these parameters in both A. seyal plantations and natural stands. The techniques and equations derived from the EWCM, utilizing L-band-like polarization data, can be employed to spatially map the distribution of aboveground biomass and carbon in dry forests.