Landscape scale carbon stock assessment of tropical peat swamp forests using an integrated field measurement and remote sensing technique : a case study in pt diamond raya timber, Rokan Hilir district, Riau province, Indonesia

Abstract

Indonesia's tropical peatland ecosystems constitute one of the largest terrestrial carbon pools in the world. However, anthropogenic and natural disturbances can convert the peatland ecosystems' status from carbon sinks and stores to sources. Huge amounts of carbon are released to the atmosphere as CO2 through peat decomposition and fire. When these emissions are taken into account, Indonesia became one of the largest CO2 emitter worldwide. Despite their significant role in the global climate change, tropical peatland ecosystems (particularly peat swamp forests) are inadequately studied. Thus, the main objectives of this research are to fill the gap in knowledge regarding peat swamp forests and to develop a practical framework for assessing carbon stock at landscape level in peat swamp forests under an active logging company. Site-specific allometric equations for above-ground biomass and carbon (AGB and AGC) were developed based on 51 and 31 destructively sampled trees, respectively, and the results were compared to the published equations commonly used in the tropical forest biomass studies. The carbon fraction used to develop AGC equations were analysed based on 362 subsamples from 31 sampled trees using the Walkley-Black method. The relationship between total tree height (H) and diameter at breast height (DBH) were constructed based on 286 sampled trees using linearized and non-linear functions. A total of 150 measurement plots were used to analyse the impacts of selective logging on forest community composition, structure and biomass. The stand attributes comparison between logged and unlogged plots was conducted using non-parametric tests. The landscape level carbon stock estimates were calculated based on a published biomass-backscatter algorithm and ALOS-PALSAR images. The results indicated that the developed site-specific allometric equations have determination coefficient of greater than 95%. The model which used DBH, H and wood density has the best performance, thus is recommended when reliable data on these variables are available. The site-specific outperformed the published equations in estimating AGB and AGC. The non-linear function provided better H-D models than the linearized function based on the statistical parameters and the biological criteria with the fit index of ~91% and the standard error of 3.6 m. The model based on the modified logistic function is recommended. The species composition, structure and biomass between selectively logged and unlogged forest plots were statistically significantly different, but there are indications that the differences decrease with time since harvest. The generated landscape level AGB map depicts logically the spatial pattern of AGB density distribution, thus is useful for management purposes, particularly in allocating the resources efficiently for further ground-based investigations. The landscape level AGB per ha ranged from 12.5 to 390.6 t with an average of 213.4 t. The carbon fraction obtained in this study was 0.473, which was slightly lower than the generic value of 0.5. Despite the limitations, this study has provided a practical framework for assessing carbon stocks and the impacts of selective logging at the landscape level.