Abstract
Integration of optical, lidar, and radar data for estimating forest structural parameters has been extensively investigated in recent decades. While evidence in the literature reveals that a common fusion method, where all attributes derived from different types of remotely-sensed data are used directly in stepwise multiple-linear regression, is not able to increase the accuracy of the structural parameter estimations in all cases, it is possible to determine the ratio of the attributes (called the pseudo-attribute hereafter) derived from two different datasets for the modelling process. In this study, the performance of WorldView-2 (WV-2) and Satellite Pour l’Observation de la Terre-5 (SPOT-5) multispectral images, small foot-print lidar data, and multi-date dual-polarized Advanced Land Observing Satellite phased array type L-band synthetic aperture radar (ALOS–PALSAR) data in paired-data fusion have been assessed using the two methods of common and ratio fusion for estimating forest structural parameters over a Pinus radiata plantation. For this purpose, grey level co-occurrence matrix (GLCM) indices with different orientations and window sizes were calculated for the WV-2 and SPOT-5 multispectral data. The backscatter derivatives and statistical metrics were extracted from multi-date dual-polarized ALOS-PALSAR data and lidar-derived canopy height model (CHM), respectively. After applying stepwise multiple-linear regression, the results showed that the ratio fusion method provided more accurate models than the common fusion method. Finally, statistical analysis showed that there is no significant difference between results derived when SPOT-5 and WV-2 data were each fused with lidar data, resulting in estimation of structural parameters with less than 20% error, which is consistent with the quality of field inventories.