Pine Forest Structural Parameter Retrieval using Radar Images

Structural parameter estimation of the forests including deciduous and coniferous is required for understanding environmental cycles including Carbone cycle, hydrological cycle and etc., in a global scale, and sustainable management of the forests, in local scale. Although, the feasibility of different remotely sensed data including optical, radar and Lidar as reliable alternatives for conventional inventory methods have been frequently used for estimating forest structural parameters, the use of multi-temporal radar data have been studied less than the other options for this purpose. In this study the radar images acquired in different dates, were utilized to estimate the structural parameters of a pine plantation. For this purpose, geometric correction and speckle noise reduction methods were applied on multi-polarized Advanced Land Observing Satellite (ALOS)-Phased Array type L-band Synthetic Aperture Radar (PALSAR) data. Afterwards, different backscatter derivatives along with their corresponding textural information were extracted using grey level co-occurrence matrix (GLCM) for different window sizes and orientations. Afterwards, a stepwise multiple-linear regression was applied to model the relationship between structural parameters and synthetic aperture radar (SAR) attributes. The results indicated that the models based on multi-date SAR data performed better than those derived from single-date SAR data. Moreover, it was shown while the estimation error of mean height is 20.7%, the other parameters were estimated with error of more than 30%. Finally, the effects of slope and tree age on the estimation accuracy of structural parameters were investigated.