Abstract:
Above ground biomass is needed for assessing forest carbon storage. Traditional allometric approaches generally gain moderate accuracies and require destructive sampling that is sometimes impractical. As an alternative to the traditional approaches, terrestrial laser scanners (TLS) gather accurate data without destroying the tree samples. Thanks to precision and speed of the TLS instrument that provide sufficient details of three dimensional point cloud. This study was divided in two parts. The first part was to preliminary test the capability of the methodology under a relatively controlled conditions. The TLS point data were collected from a test plot (a eucalyptus plantation), and three surface reconstruction algorithms, Quantitative structure models (QSM), Poisson surface reconstruction (PSR), and Screen Poisson surface reconstruction (SPSR) respectively, were applied to build volumetric models of sample trees. It is notable that this is the first study to test the SPSR method on real tree samples. The results were then compared with the reference values measured by a water replacement method. The SPSR approach yielded the most accurate results (RMSE 0.64 kg. or 7.02%). While the PSR method gained an RMSE of 0.76 kg. (8.41%). The QSM method had the worst results with an RMSE of 1.36 kg. (14.25%). The second part of the study was to apply the proposed methodology to a real world scenario. The TLS instrument was used for collecting the data from an Avicennia marina mangrove conservation area. Since destructive sampling was not allowed in the study area, the results of the TLS method were compared to five published allometric models without cutting any tree sample. The comparison showed that the above-ground biomass calculated from the three reconstruction approaches was rather different from the generic mangrove model. The mean RMSE was as large as 69.22 kg. The situation was improved when comparing the results with a species specific model, the Avicennia marina mangrove model. The mean RMSE was changed to 43.98 kg. The cause of the errors was mainly the occlusion at the upper section of the tree. It is suggested that the problem can be solved by increasing the height of the reference targets. The outcome of this work confirms the potential of the TLS instrument for AGB estimation in both natural and man-made scenarios.