- Open Access
Time-series maps of aboveground carbon stocks in the forests of central Sumatra
© Thapa et al. 2015
- Received: 24 April 2015
- Accepted: 2 September 2015
- Published: 17 September 2015
Efforts to reduce emissions from deforestation and forest degradation in tropical Asia require accurate high-resolution mapping of forest carbon stocks and predictions of their likely future variation. Here we combine radar and LiDAR with field measurements to create a high-resolution aboveground forest carbon stock (AFCS) map and use spatial modeling to present probable future AFCS changes for the Riau province of central Sumatra.
Our map provides spatially explicit estimates of the AFCS with an accuracy of ±23.5 Mg C ha−1. According to this map, the natural forests in the province currently store 265 million Mg C, with a density of 72 Mg C ha−1, as aboveground biomass. Using a spatially explicit modeling technique we derived time-series AFCS maps up to the year 2030 under three forest policy scenarios: business as usual, conservation, and concession. The spatial patterns of AFCS and their trends under different scenarios vary on a local scale, and some areas are highlighted that are at eminent risk of carbon emission. Based on the business as usual scenario, the current AFCS could decrease by 75 %, which may lead to the release of 747 million Mg CO2. The other two scenarios, conservation and concession, suggest the risk reductions by 11 and 59 %, respectively.
The time-series AFCS maps provide spatially explicit scenarios of changes in AFCS. These data may aid in planning Reducing Emissions from Deforestation and forest Degradation in developing countries projects in the study area, and stimulate the development of AFCS maps for other regions of tropical Asia.
- Forest carbon
- Aboveground biomass
Across the world, existing tropical forest landscapes are undergoing rapid deforestation due to natural disasters, as well as the increasing demand for agricultural land, wood products, energy, and developmental projects. Currently, global forest areas account for 3.85 billion ha, or 26 % of the Earth’s land surface , but this area is decreasing at around 13 million ha per year . As deforestation continues, the Earth becomes more susceptible to potentially negative impacts on ecosystems and the overall climate system due to the associated effects on carbon balance, biodiversity, soil, water regulation, and weather patterns. Currently, emissions caused by deforestation worldwide are considered to be very high, and are likely to continue in this way for the coming decades. Tropical forest regions in particular are major potential sources of carbon emissions [3–7]. Reducing Emissions from Deforestation and forest Degradation in developing countries (REDD+)  is one of the key global initiatives that aims to conserve forests and reduce carbon emissions. The goal of REDD+ is to connect investors to forest users and offer an economic portfolio for the retention of forest carbon and the avoidance of deforestation, while also slowing the drivers of land use change. As a result, the initiative contributes indirectly to biodiversity conservation by helping to reduce habitat loss and ensure the continuation of normal ecosystem services; hence, it is considered a sustainable option for the maintenance of forests. Meaningful implementation of REDD+ requires accurate, high-resolution, spatially explicit maps of forested areas and forest carbon stocks, as well as predictions of their change in the future. Therefore, efforts to improve the methods for mapping forest extents and forest-related carbon stocks, as well as identifying their changes, have been advancing in many parts of the world, including tropical Asia [5, 9–13]. Remote sensing and spatial modeling techniques offer a practical means to monitor and examine changes in forest cover, analyze the implications of forest policies, predict spatial patterns of forest cover in the future, and relate these patterns to carbon stock densities [9, 14–16].
Accurate mapping of aboveground forest carbon stocks (AFCS) using spaceborne satellite data is still very challenging due to the requirement of a large amount of in situ data for forest carbon estimation model calibration and validation. Although traditional plot-based field measurements of AFCS have proven most accurate, they are costly and difficult to implement for large areas with dense tropical forests. Studies [13, 14, 17] have shown that light detection and ranging (LiDAR) techniques allow the accurate measurements of geographically referenced vertical forest structures, including canopy height, volume, and biomass. Using LiDAR data, an allometric model for AFCS can be developed with a relatively small number of field measurements [13, 17]. Modeling results can be used to extend the field data, providing spatially extensive and detailed forest attribute data, that can be used to calibrate AFCS predictive models build around a wide variety of spaceborne data, including synthetic aperture radar (SAR) and optical imageries covering larger areas [14, 18].
The integration of both airborne and spaceborne remote sensing techniques has offered the opportunity to more precisely map forest cover and related carbon stocks over wider areas at suitable spatiotemporal scales [6, 13, 14, 19]. However, the potential application of optical spaceborne remote sensing data in Asian tropical forest regions is limited, due to the frequent appearance of clouds and haze, as well as the insensitivity of sensing systems to the variability of biomass with a multi-layer canopy in highly dense forests. In contrast, spaceborne SAR is not limited by these factors as it penetrates clouds to image the Earth’s surface regardless of weather conditions or solar illumination. Among the available spaceborne SAR systems, the Advanced Land Observation Satellite (ALOS) Phase Arrayed L-band SAR (PALSAR) operating at a wavelength of 23.6 cm is very sensitive to forest structure, yielding valuable information for the mapping of forest cover [20–22] and AFCS measurements [23, 24]. However, studies have shown that saturation remains a dominant issue when directly estimating AFCS using SAR data in high biomass areas [25–29]. A consideration of multi-temporal SAR data with multiple polarizations and the use of rule-based algorithms can help to mitigate the saturation problem and improve AFCS estimations [14, 24].
Field measurements, LiDAR data, time-series PALSAR data, and a rule-based algorithm were used together to create a baseline AFCS map with high spatial resolution. The spatial model developed by Thapa et al.  was applied to visualize and assess the implications of different forest policies on future AFCS.
Summary of field measurement plots and AFCS estimates by forest types
Field plots in %
AFCS (Mg C ha−1)
The maximum likelihood algorithm (MLA) mapping procedure was used to create four maps using the PALSAR gamma-naught image from 2010, three AFCS maps to which the Lee, Frost, and median filters had been applied, respectively, with a 3 × 3 window, and an AFCS map without any filters for comparison. Validation was performed with three individual statistical measures to assess the filtering effects. The median filter provided the best AFCS map, with a RMSE and bias of 27.59 and −0.83 Mg C ha−1 and an index of agreement (D) of 0.74. The other two filters, Lee and Frost, gave RMSEs with biases of 28.03 and −1.28, and 27.47 and −3.67 Mg C ha−1, and D values of 0.71 and 0.74, respectively. The map created without the application of any filters had an RMSE and bias of 30.09 and −2.71 Mg C ha−1, respectively, and a D of 0.59. The D values are similar in both the Frost and the median filters, although the RMSE indicates a slightly better performance with the Frost filtered map, the bias is comparatively high. Overall, these results suggest that the median filter provides a good AFCS mapping product.
The level of error in this map may have resulted from several unquantifiable factors, including differences in field measurement processes, inaccuracies in the field and LiDAR allometric equations, the slope correction method for the PALSAR mosaic data, and the time difference between field and remote sensing measurements . However, the mapping error produced herein is low, and to date remains unmatched in similar studies of tropical forest regions, such as Kalimantan , the Amazon , and other areas . Owing to the low mapping error and the high level of similarity between predicted and observed AFCS in such a diverse tropical forest, this map was used as a baseline for a spatial model that calculates future changes in the forest carbon footprint of the study area.
Quantity of AFCSs extracted for natural forest areas and their distribution by district
Forest area in ha
In million Mg C
Density in Mg C ha−1
Despite the high spatial resolution of our AFCS map, which reveals the AFCS dynamics of natural forests on a local scale, information about the dynamics of below-ground carbon stocks (BGCS) in natural forest areas is lacking in this study. Estimation of BGCS in natural forest regions using remote sensing is extremely difficult. However, the general model [32, 33] for estimating the BGCS in tropical regions can be employed using the baseline map derived in this study if necessary. In addition to affecting the BGCS, the deforestation process also contributes to the release of carbon from other sources such as soils and peat lands. These additional sources may significantly increase the net carbon emissions, and the inclusion of forest fire parameters in the modeling process may improve the accuracy of future estimation. Furthermore, the majority of the natural forest land in this area has already undergone transformations to produce economically valuable industrial plantations, such as oil palm, acacia, coconut, and rubber trees. These plantation forests also store a significant amount of aboveground forest carbon, as reflected in baseline map (Fig. 3) and in Table 1. From a carbon stock perspective, a consideration of the AFCS dynamics in these forests may represent a trade off in overall carbon balance to some extent. However, there is an immediate risk of carbon emission when the natural forests are cleared, even when they are replaced by plantation forests. Additionally, these plantation forests store carbon only in the short term, as they are harvested within a certain period of time. For example, an acacia plantation in the study area is harvested every 4 years. The other plantations, such as oil palm, coconut, and rubber are often harvested every 20–30 years. The plantation forests maintain greenness in the province but still have a significant impact on carbon recycling and prevent the restoration of ecosystem services.
Through the integration of multiple remote sensing techniques, from airborne LiDAR to spaceborne SAR with field measurement data and a rule-based algorithm, an accurate baseline AFCS map with high spatial resolution was developed for one of the major tropical forests in Asia. This baseline map provides highly accurate, spatially explicit distributions and quantitative estimations of forest carbon stocks in the study area. The AFCS distribution varies geographically, indicating spatial variations in forest quality and vulnerability. The spatial modeling technique provides an opportunity to extrapolate the spatial trends in AFCS and examine the implications of different forest management policies on carbon stocks and emissions over the next two decades. The inherent capability of the model to distinguish local variations in future AFCS trends under different scenarios is key to identifying the areas most vulnerable to high carbon emissions, which would require immediate mitigation measures to ensure forest conservation. The model was used to predict the spatiotemporal variations and associated quantities of remaining AFCS under different scenarios up until the year 2030. These predicted spatial patterns of AFCS indicate that the forest carbon emission rate is likely to be high in the coming decades across the province. Ongoing deforestation is expected to release around 747 million Mg CO2 into the atmosphere by 2030. A forest conservation policy will slow the AFCS emissions, but the reduction will be insufficient. Among the scenarios tested, the concession scenario is the most promising, halving the expected emissions if it is implemented as planned. In addition to the high-resolution AFCS map, the modeling outcomes may provide opportunities for the identification of especially vulnerable localities and focuses for the implementation of REDD+ projects to obtain the greatest benefits based on the environmental settings. For the proper execution of the REDD+ project, it is important to understand how the expected trends in AFCS distribution are likely be affected in the long term by the implementation various plans, policies, and strategies. The AFCS under the BAU scenario may provide a reference emission scenario for REDD+, while the other scenarios can be used as examples in the initial exploration of the range of potential spatiotemporal issues and outcomes. These can provide important insights for preparedness activities that mitigate the problem of forest carbon emissions. The spatially explicit AFCS map and the modeled scenario results will therefore contribute to the sustainable management of forests in the study area and to the formulation of REDD+ projects, as well as representing a methodological reference for wider audiences in tropical regions and beyond.
L-band SAR data collection and processing
The study area spans more than 9 million ha commonly experiences cloudy and hazy skies throughout the year . This degradation in atmospheric conditions over the study area precludes use of optical remote sensing techniques in assessing forest quality and AFCS. As a result, PALSAR mosaic data used for tropical forest monitoring as they are unaffected by atmospheric condition and are available for wall-to-wall mapping [1, 20]. The mosaic data are slope-corrected and orthorectified using the widely available SRTM 90 m digital elevation model without any alteration the image quality . Currently, 25 m global mosaic data in two polarizations (HH and HV) are available as one set per year, from 2007 to 2010. The mosaic data are available at a downloadable size of 1-degree tiles, equivalent to approximately 111 × 111 km . Those mosaic products covering the whole province of Riau for the years 2009 and 2010 were used. The mosaic data were converted into radar backscatter coefficients using gamma-naught (γ°)  due to high sensitivity to forest structure and its usefulness in forest cover analysis [20, 35].
To improve the confidence of the AFCS mapping, we also examined whether a particular filter or its size would affect the mapping results. Three filters were examined: the Lee, Frost, and median filters. These filters possess different formulations and assumptions for smooth speckled data in radar imagery. The Lee filter is a standard deviation-based filter, and filters data on the basis of statistics calculated within individual filtering windows. It conserves image sharpness and details while reducing speckle noise. The value of the pixel being filtered is replaced by a value computed using the neighboring pixels. In comparison, the Frost filter is an exponentially damped circularly symmetric filter, which utilizes local statistics. The value of the pixel being filtered is replaced by a value computed with a consideration of the damping factor, the local variance, and the distance from the filter center. This filter is able to preserve the edges in the images. The median filter reduces the speckle noise in an image by conserving edges greater than the kernel dimensions. It replaces the value of each center pixel with the median value of the neighborhood specified by the filter size. In order to investigate the impact of speckle filtering sizes on the mapping, we also evaluated five different filtering window sizes (3 × 3, 5 × 5, 7 × 7, 9 × 9, and 11 × 11).
Field data collection and processing
The combination of LiDAR data and plot-based field measurements has emerged as a promising technique for accurately estimating AFCS [13, 17]. We conducted field measurements and airborne LiDAR surveys within the province during 2012 and 2013. Owing to the differences in forest structure and associated biomass in different land use and land cover (LULC) types, we adopted a stratified sampling approach based on the major forest types to determine the locations (Fig. 1) for field and LiDAR measurements. The major forest types in the study site were defined as natural forests, including peat swamps, dry moist, mangrove, and regrowth, and plantations including acacia, oil palm, rubber, and coconut. Based on these forest types, eight strata were created.
Across the field measurement campaigns, we made 87 biomass measurements within 1 ha-size plots that coincided with the LiDAR acquisition sites. Forest stands of all ages were inventoried, from mature to recent regrowth. Owing to the time and cost involved in conducting a census-based measurement of all trees in a 1-ha-sized plot, a sub-sampling approach was adopted using representative subplots. The sub-sampling methods differed between the natural and plantation forests.
To determine woody biomass, all living and standing deadwood trees with a diameter at breast height (at 1.3 m; DBH) ≥5 cm were measured in each subplot. We used allometric equations previously developed for the specific forest types: peat swamp forest , dry moist forest , mangrove , acacia , rubber , coconut and non-trees , oil palm , standing deadwood , lying deadwood , and bamboo . The biomass of understory vegetation and litter was calculated by multiplying the mass of a fresh sample measured in the field by the ratio of sub-sample dry mass to sub-sample fresh mass. The plots within regrowth forests were all located in either peat swamps or dry moist forested areas. Therefore, the biomass for this class was calculated using the corresponding allometric equation, based on its location. Detailed descriptions of the field measurement method, plot specification, and allometric equations used to convert the field measured data to plot level aboveground biomass (AGB) are presented in Thapa et al. [13, 24]. In this study, the AFCS for each plot was considered to be 47 % of the field measured AGB .
A total of 2716 1-ha LiDAR-based AFCS plots were created, avoiding an excess of path boundaries, field measurement plots, agricultural fields, clear cut areas, water areas, and built structures including buildings and roads. These LiDAR AFCS plots were combined with the 87 field measurement plots, resulting in a total of 2803 plots available for calibration and validation of the SAR-based AFCS baseline map.
Mapping and validating the AFCS
Mapping of future expected AFCS footprints
In this study, we used the forest cover map derived from PALSAR mosaic data from 2010, in combination with the scenario maps for 2015, 2020, 2025, and 2030 from Thapa et al.  to generate the expected AFCS footprints for the various years in the future. Three policy scenarios were analyzed: BAU, corresponding to the ‘business as usual policy’, G-FC indicating the ‘government-forest conservation policy’, and G-CPL, representing the ‘government-concession for plantations and logging policy’. The BAU policy scenario assumes that the deforestation process will continue with the same past trend everywhere in the province, and therefore, AFCS removal will occur in the corresponding deforested areas. The G-FC policy scenario assumes that the deforestation process does not follow the past trend and, in the future, will likely occur outside the designated forest conservation areas. In this case, the forest carbon stocks remain untouched within the conservation areas. For the G-CPL policy scenario, we assume that the concession areas are allotted for selective logging and industrial plantations, and imminent deforestation likely occurs only in the concession areas, therefore the AFCS will be untouched outside these areas. Scenario-wide AFCS maps were created at five-year intervals from 2015 to 2030. Using these maps, the AFCS was quantified at the district level to identify local variation in the carbon stocks. Furthermore, the expected CO2 emissions for each scenario were computed at province level for each time interval using the approach of IPCC .
RBT, TM, MW, and MS conceived the research and analyzed the PALSAR data. RBT, TM and MW collected, processed, and analyzed the field data. RBT created the forest carbon map, performed spatial modeling, and wrote the manuscript. MS arranged necessary funds and coordinated the local collaborator for the research. All authors read and approved the final manuscript.
The authors would like to thank Prof. I. N. S. Jaya, E. S. Purnama and other colleagues and their students from IPB, Indonesia, Mr. K. Shono and his colleagues from Hatfield Consultants, Indonesia, Mr. Tomohiro Shiraishi from JAXA, and Mr. Takuya Itoh from RESTEC for their assistance with field data collection and for sharing their thoughts in this area of research.
Compliance with ethical guidelines
Competing interests The authors declare that they have no competing interests.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
- Shimada M, Itoh T, Motohka T, Watanabe M, Shiraishi T, Thapa R, Lucas R. New global forest/non-forest maps from ALOS PALSAR data (2007–2010). Remote Sens Environ. 2014;155:13–31.View ArticleGoogle Scholar
- FAO. Forest resource assessment 2010. Rome: Food and Agricultural Organization of the United Nations; 2012.Google Scholar
- Aguilar-Amuchastegui N, Riveros JC, Forrest JL. Identifying areas of deforestation risk for REDD+ using a species modeling tool. Carbon Bal Manag. 2014;9:10.View ArticleGoogle Scholar
- Canadell JG, Schulze ED. Global potential of biospheric carbon management for climate mitigation. Nat Commun. 2014;5:5282.View ArticleGoogle Scholar
- Grace J, Mitchard E, Gloor E. Perturbations in the carbon budget of the tropics. Glob Change Biol. 2014;20:3238–55.View ArticleGoogle Scholar
- Harris NL, Brown S, Hagen SC, Saatchi SS, Petrova S, Salas W, Hansen MC, Potapov PV, Lotsch A. Baseline map of carbon emissions from deforestation in tropical regions. Science. 2012;336:1573–6.View ArticleGoogle Scholar
- Strassburg B, Turner KR, Fisher B, Schaeffer R, Lovtt A. Reducing emissions from deforestation—The combined incentives mechanism and empirical simulations. Glob Environ Chang. 2009;19:265–78.View ArticleGoogle Scholar
- http://www.un-redd.org. Accessed 29 Sept 2014.
- Carlson KM, Curran LM, Ratnasari D, Pittman AM, et al. Committed carbon emissions, deforestation, and community land conversion from oil palm plantation expansion in West Kalimantan, Indonesia. Proc Natl Acad Sci USA. 2012;109:7559–64.View ArticleGoogle Scholar
- Cutler M, Boyd D, Foody G, Vetrivel A. Estimating tropical forest biomass with a combination of SAR image texture and Landsat TM data: an assessment of predictions between regions. ISPRS J Photogram. 2012;70:66–77.View ArticleGoogle Scholar
- Leach M, Scoones I. Carbon forestry in West Africa: the politics of models, measures and verification processes. Glob Environ Chang. 2013;23:957–67.View ArticleGoogle Scholar
- Morel AC, Fisher JB, Malhi Y. Evaluating the potential to monitor aboveground biomass in forest and oil palm in Sabah, Malaysia, for 2000–2008 with Landsat ETM+ and ALOS-PALSAR. Int J Remote Sens. 2012;33:3614–39.View ArticleGoogle Scholar
- Thapa RB, Watanabe M, Motohka T, Shiraishi T, Shimada M. Calibration of aboveground forest carbon stock models for major tropical forests in central Sumatra using airborne LiDAR and field measurement data. IEEE J Sel Top Appl. 2015;8:661–73.Google Scholar
- Asner GP, Knapp DE, Martin RE, Tupayachi R, Anderson CB, Mascaro J, Sinca F, Chadwick KD, Higgins M, Farfan W, Llactayo W, Silman MR. Targeted carbon conservation at national scales with high-resolution monitoring. Proc Natl Acad Sci USA. 2014;111:E5016–22.View ArticleGoogle Scholar
- Lin L, Sills E, Cheshire H. Targeting areas for reducing emissions from deforestation and forest degradation (REDD+) projects in Tanzania. Glob Environ Chang. 2014;24:277–86.View ArticleGoogle Scholar
- Thapa RB, Shimada M, Watanabe M, Motohka T, Shiraishi T. The tropical forest in South East Asia: monitoring and scenario modeling using Synthetic Aperture Radar data. Appl Geogr. 2013;41:168–78.View ArticleGoogle Scholar
- Asner GP, Powell GVN, Mascaro J, Knapp DE, Clark JK, Jacobson J, Kennedy-Bowdoin T, Balaji A, Paez-Acosta G, Victoria E, Secada L, Valqui M, Hughes RF. High-resolution forest carbon stocks and emissions in the Amazon. Proc Natl Acad Sci USA. 2010;107:16738–42.View ArticleGoogle Scholar
- Tsui OW, Coops NC, Wulder MA, Marshall PL. Integrating airborne LiDAR and space-borne radar via multivariate kriging to estimate above-ground biomass. Remote Sens Environ. 2013;139:340–52.View ArticleGoogle Scholar
- Mitchard ETA, Saatchi SS, Baccini A, Asner GP, Goetz SJ, Harris NL, Brown S. Uncertainty in the spatial distribution of tropical forest biomass: a comparison of pan-tropical maps. Carbon Bal Manag. 2013;8:10.View ArticleGoogle Scholar
- Thapa RB, Itoh T, Shimada M, Watanabe M, Motohka T, Shiraishi T. Evaluation of ALOS PALSAR sensitivity for characterizing natural forest cover in wider tropical areas. Remote Sens Environ. 2014;155:32–41.View ArticleGoogle Scholar
- Shimada M, Ohtaki T. Generating large-scale high-quality SAR mosaic datasets: application to PALSAR data for global monitoring. IEEE J Sel Top Appl. 2010;3:637–56.Google Scholar
- Motohka T, Shimada M, Uryu Y, Setiabudi B. Using time series PALSAR gamma nought mosaics for automatic detection of tropical deforestation: a test study in Riau, Indonesia. Remote Sens Environ. 2014;155:79–88.View ArticleGoogle Scholar
- Saatchi S, Marlier M, Chazdon RL, Clark DB, Russell AE. Impact of spatial variability of tropical forest structure on radar estimation of aboveground biomass. Remote Sens Environ. 2011;115:2836–49.View ArticleGoogle Scholar
- Thapa RB, Watanabe M, Motohka T, Shimada M. Potential of high-resolution ALOS-PALSAR mosaic texture for aboveground forest carbon tracking in tropical region. Remote Sens Environ. 2015;160:122–33.View ArticleGoogle Scholar
- Dobson MC, Ulaby FT, Le Toan T, Beaudoin A, Kasischke ES, Christensen NC. Dependence of radar backscatter on conifer forest biomass. IEEE Trans Geosci Remote. 1992;30:412–5.View ArticleGoogle Scholar
- Lucas R, Armston J, Fairfax R, Fensham R, Accad A, Carreiras J, Kelley J, Bunting P, Clewley D, Bray S, Metcalfe D, Dwyer J, Bowen M, Eyre T, Laidlaw M, Shimada M. An evaluation of the ALOS PALSAR L-band backscatter—above ground biomass relationship Queensland, Australia: impact of surface moisture condition and vegetation structure. IEEE J Sel Top Appl. 2010;3:576–93.Google Scholar
- Englhart S, Keuck V, Siegert F. Aboveground biomass retrieval in tropical forests—the potential of combined X- and L-band SAR data use. Remote Sens Environ. 2011;115:1260–71.View ArticleGoogle Scholar
- Luckman AJ, Baker JR, Honzák MH, Lucas RM. Tropical forest biomass density estimation using JERS-1 SAR: seasonal variation, confidence limits and application to image mosaics. Remote Sens Environ. 1998;62:126–39.View ArticleGoogle Scholar
- Watanabe M, Motohka T, Shiraishi T, Thapa RB, Kawano N, Shimada M. Dependency of forest biomass on full polarimetric parameters obtained from L-band SAR data for a natural forest in Indonesia. In: Proc Intl Geosci Remote Sens Symp 2013, 3919–3922.Google Scholar
- MofF. IFCA consolidation report: reducing emissions from deforestation and forest degradation in Indonesia. Forestry Research and Development Agency, Jakarta, Indonesia; 2008.Google Scholar
- BPS. Hasil Sensus Pnduduk 2010: data Agregat per Provinsi (in Indonesian language); 2010. Web: http://dds.bps.go.id/eng/download_file/SP2010_agregat_data_perProvinsi.pdf. Accessed 2012.3.7.
- Cairns MA, Brown S, Helmer EH, Baumgardner GA. Root biomass allocation in the world’s upland forests. Oecologia. 1997;111:1–11.View ArticleGoogle Scholar
- Pearson T, Walker S, Brown S. Sourcebook for land use, land-use change and forestry projects. Little Rock: Biocarbon Fund, Winrock International; 2005.Google Scholar
- http://www.eorc.jaxa.jp/ALOS/en/dataset/dataset_index.htm. Accessed in 2015.6.30.
- Shiraishi T, Motohka T, Thapa RB, Watanabe M, Shimada M. Comparative assessment of supervised classifiers for land use land cover classification in a tropical region using time-series PALSAR mosaic data. IEEE J Sel Top Appl. 2014;7:1186–99.Google Scholar
- Murdiyarso D, Rosalina U, Hairiah K, Muslihat L, Suryadiputra INN, Jaya A. Petunjuk Lapangan: Pendugaan Cadangan Karbon pada Lahan Gambut. Proyek Climate Change, Forests and Peatlands in Indonesia. Wetlands International, Indonesia Programme and Wildlife Habitat Canada. Bogor, Indonesia; 2004.Google Scholar
- Brown S. Estimating biomass and biomass change of tropical forests: a primer. UN FAO Forestry Paper, no. 134; 1997. p 55.Google Scholar
- Komiyama A, Poungparn S, Kato S. Common allometric equations for estimating the tree weight of mangroves. J Trop Ecol. 2005;21:471–7.View ArticleGoogle Scholar
- Adiriono T. Measurement of carbon stock with carbonization method in forest plantation of acacia crassicarpa: a case study in PT Sebangun Bumi Andalas Wood Based Industries. MSc Thesis, Faculty of Forestry, Gadjah Mada University; 2009.Google Scholar
- Schroth G, Angelo SAD, Teixeira WG, Haag D, Lieberei R. Conversion of secondary forest to agroforestry and monoculture plantations in Amazonia: consequences for biomass, litter and soil carbon stock after 7 years. For Ecol Manag. 2002;163:131–50.View ArticleGoogle Scholar
- Frangi JL, Lugo AE. Ecosystem dynamics of a subtropical floodplain forest. Ecol Monogr. 1985;55:351–69.View ArticleGoogle Scholar
- Yulianti N. Carbon stock of peatland in oil palm agroecosystem of PTPN IV Ajamu, Labuhan Batu, North Sumatra. MSc Thesis, Faculty of Agriculture, Bogor Agricultural University (IPB), Bogor; 2009.Google Scholar
- Walker SM, Pearson TRH, Casarim FM, Harris N, Petrova S, Grais A, Swails E, Netzer M, Goslee KM, Brown S. Standard operating procedures for terrestrial carbon measurement. Little Rock: Winrock International; 2012.Google Scholar
- Pearson T, Walker S, Brown S. Sourcebook for land use, land-use change and forestry (LULUCF) projects. Little Rock: BioCarbon Fund and Winrock International; 2005.Google Scholar
- van Noordwijk M, Rahayu S, Hairiah K, Wulan YC, Farida A, Verbist B. Carbon stock assessment for a forest-to-coffee conversion landscape in Sumber-Jaya (Lampung, Indonesia): from allometric equations to land use change analysis. Sci China (Ser C). 2002;10:75–86.Google Scholar
- IPCC. Agriculture, forestry and other land use. Guidelines for National Greenhouse Gas Inventories, 4; 2006. http://www.ipcc-nggip.iges.or.jp/public/2006gl/vol4.html. Accessed 2 Feb 2012.
- Richards JA, Jia X. Remote sensing digital image analysis. Berlin: Springer; 2006.Google Scholar
- Willmott CJ, Wicks DE. An empirical method for the spatial interpolation of monthly precipitation within California. Phys Geogr. 1980;1:59–73.Google Scholar