Effect of topographic aspect and polarization in estimation of some forest variables using ALOS-2 data (Case study: Shastkalateh forest, Gorgan)

SAR data have been widely used for estimation of forest variables. These data have a high potential in flat terrain, however, there are some problems such as the effect of topography on backscatters in mountainous regions. The goal of this study was to evaluate the effect of oriented and non-oriented aspect slopes and polarization in the estimation of stand Lorey’s mean tree height, volume, and basal area variables using dual-polarization (HH, HV) data of ALOS-2 PALSAR-2 in a part of Shastkalateh forest of Gorgan. Modeling was performed in three modes including without considering the aspect slope, in the slope-aspect orientated with the transmitted waves and non-orientated with waves, and also in the polarizations as separately using linear regression and support vector machine methods. The results of correlation between the backscattering of the PALSAR-2 and studied variables showed that in the slope-aspect orientated with the transmittance, the correlations were significant for basal area and volume, but it was not significant for Lorey's height. While in the non-orientated slopes, the correlation rate was very low for all variables and no significant. Also, the results showed that the HV polarization and the support vector machine method were very sensitive to the Lorey’s mean tree height, basal area, and volume variables with r and relative RMSE (-0.14, 12.44), (-0.48, 29.35) and (-0.44, 36.40), in the area in the direction of alignment with the waves, respectively. While the lower sensitivity is (-0.18, 14.14), (-0.05, 35.85), and (-0.04, 38.40), respectively, in the non-orientated with waves.