Simulation Impact of REDD Policy: Case Study of Forest Area in Indonesia

Indonesia's forests in different periods have been deforested at different levels. Deforestation caused carbon emissions. The purposes of this study were :1) to measure deforestation and carbon emissions in period of 2005-2010 in Indonesia and 2) to find out the incentive value to be paid by the government. One method for measuring emissions from deforestation and forest degradation is GeOSIRIS model. A modeled GeOSIRIS policy used a carbon payment system to incentivize emission reductions. Data used in this study were maps of forest cover in 2005 and 2010, map of deforestation 2005-2010, carbon and agricultural price and driver variables for deforestation such as slope, elevation, logarithmic distance to the nearest road, logarithmic distance to the nearest provincial capital, the amount of area per pixel included in a national park, a timber plantation. The result of this study showed rate of deforestation was 4.65 million ha/5 years. The REDD policy could decrease deforestation in Indonesia by 0.66 million ha (17.45 %). Assuming that International carbon price was US$ 10/tCO2e, the change of emissions due to REDD was 24.75%, or reduced emissions by 1.09 million tCO2e/5 years. Finally, Gross National Revenue from carbon payments (NPV 5 years) was US$ 10.917 billion, where incentivize emission reductions to sub-national entities (NPV, 5 years) was US$ 9.178 billion and net central government surplus from carbon payments was US$ 1.739 billion (NPV, 5 years).


I. INTRODUCTION
Tropical forests and other vegetated landscapes like grasslands and wooded savannahs play a major role in the global carbon sequestration process and their conservation and protection offers immense potential for reducing greenhouse gas emissions and global warming [5]. Referring to [3] that clearing of primary forests also results in the destruction of unique tropical forest habitats, thus causing the loss of biodiversity. Among tropical countries Indonesia experiences the second highest rate of deforestation. Therefore, accurate and up-to-date forest data are required to fight deforestation and forest degradation to support initiatives of climate change mitigation and biodiversity conservation policy [8]. Meanwhile [16] explained that the largest deforestation in Indonesia occurred in Kalimantan and Sumatra with a percentage of 36.32% and 24.49% respectively, followed by Sulawesi 11.00%, Java 9.12%, Maluku 8.30%, Bali-Nusa Tenggara 6.62%. Papua became the smallest area contributing to deforestation of 4.15%. It could be seen that deforestation in Indonesia until 2009 was concentrated in Kalimantan and Sumatra. Out of the 15.79 Mha of forest cover loss in Indonesia, reported 38% (6.02 Mha) happened inside primary intact or damaged forests [10]. Meanwhile [11] said that over the study period annual primary forest cover loss increased with the highest total loss happened in 2012 (0.84Mha). The number was greater than the reported forest loss in Brazil (0.46Mha), which was the historical leader in the tropical forest clearing. Referring to [13], Borneo Island in the period 2000-2011 has deforestation amounted to 3.040 million ha, namely deforestation in peatland forests of 0.560 million (18.42%) and deforestation in mineral land (non-peatland) for 2,480 million (81.58%). Based on the period of time of deforestation, 48.5 % of deforestation occurred in the period 2006-2011, i.e. deforestation on peatland forests of 0.334 million ha (59.69%) and deforestation in mineral forests of 1.144 million (46.15%). In Indonesia deforestation is usually linked with production of timber and expansion of settlement and agricultural area. When this existing trend continues without implementing any corrective measures, it is projected to result in a reduction of forest cover by 15% between 2015 and 2030, going from approximately 88,000,000 ha to 74,994,100 ha. On average, 830,000 ha of forest would be cleared for timber extraction or land conversion every year between 2015 and 2030. When the forest cover declines, so does the amount of carbon stored. The cumulative emissions from 2015 to 2030 due to forest loss would reach 2.5 billion tCO2, which, assuming an average carbon price of USD 5 to USD 10 per ton (based on international average market prices), would translate in a cumulative loss of about USD 10 billion to USD 25 billion between 2015 and 2030 [4]. REDD is not directed at stopping planned conversion of forests to other economic uses, nor at stopping the use of forests for timber. REDD signifies a way to value natural resource of carbon so that it can be considered along with other regular forest assets, when making decisions about land use and forest use [14]. In the calculation and modeling for carbon emissions, there are several methods and approaches. One model is the GeOSIRIS model developed by Jonah Busch at Conservation International. The GeOSIRIS model was originally developed as OSIRIS as a transparent decision support tool for REDD+ policy makers [7]. The GeOSIRIS modeler is different from the REDD modeler found in Land Change Modeler (LCM). The REDD modeler in LCM predicts how carbon emissions and deforestation would change if a certain reference area were shielded from deforestation. Meanwhile, the GeOSIRIS modeler adopts an alternate strategy. A carbon payment system is used by a modeled GeOSIRIS policy to give incentives to emission reductions. The policy can be governed at various administrative levels, such as province or district. Rather than defending a specific section of land from deforestation, scope of work for GeOSIRIS projects would be on regional or national scale, by setting a certain price to every ton of carbon dioxide emitted ($/tCO2e). The GeOSIRIS model assumes forest users encounter a trade-off between the carbon revenue obtained by protecting the forests and the agricultural revenue obtained from deforesting the land. Given some variables such as a proposed carbon price and maps of previous deforestation, the model predicts how carbon emissions, deforestation, and agricultural and carbon revenues would change if such policy were implemented [7].
The model designs balance incentives to lower usually high deforestation emissions with incentives to keep usually low deforestation emissions. Approximations of emission reductions under REDD depend significantly on the degree to which demand for tropical agriculture in the borderline generates leakage. This emphasizes the potential importance to REDD of balancing strategies to supply agricultural needs outside the forest borderline [6]. The purposes of this study were to measure deforestation and carbon emissions in period of 2005-2010 in Indonesia and to find out the incentive value to be paid by the government.

II. MATERIAL AND METHODS 2.1 Data used
This study used data from https://clarklabs.org/download/terrset-tutorial-data/, accessed on April 4, 2017, consisting of: (a) forest cover maps in 2005 and 2010, deforestation map 2005-2010 (see figure 1); (b) map of potential driver variables for deforestation, consisting of maps: slope, elevation, logarithmic distance to the nearest road, distance from the provincial capital, national park map, and plantation area map. These data are global data with spatial resolution of 3 km x 3 km. These data include global data that can be used for monitoring a large area (such as the whole Indonesia), due to the availability of sufficient data. However, for more specific planning, medium and detail scale data are needed to obtain more accurate results. The disadvantage of these data is that the spatial resolution is too small (where one pixel represents an area of 900 ha). Therefore, areas with less than 900 ha (one pixel) will be combined into a more dominant class. The map actually covered the entire territory of Indonesia. For this study other than covering Indonesia, it was also cropped to cover Kalimantan and Sumatra Islands. The GeOSIRIS model in REDD impact calculations is based on an enhanced OSIRIS model [7]. The flow chart of the GeOSIRIS modeling stage is presented in figure 2. In general, GeOSIRIS model has two main steps: (1) regression analysis, where the regression coefficient(s) and Effective Opportunity Cost image are calculated, and (2) calculations of proportional national change in agricultural price, output images (deforestation and emission), output image on administrative level decisions then the summary Excel spreadsheet is generated.

Regression Analysis
Stage of activity in this research refers to Eastman [9]. The regression step of the GeOSIRIS modeler calculates the correlation between deforestation and some individual variables (14 variables), including agricultural revenue. There are several options to classify this regression, where GeOSIRIS will run a separate regression for several different classes. These classes can be based on the amount of preexisting forest cover or geographic regions, such as provinces or districts (for geographic stratification). This study is based on geographic regions, for Indonesia such as provinces (33 provinces) or districts (426 districts), For Sumatra Island such as provinces (13 provinces) or districts (131 districts) and for Kalimantan Island, such as provinces (5 provinces) or districts (55 districts).
The regression model used in this study is Poisson regression, in which the deforestation is counted by assuming that each pixel is composed of smaller subsections which may be individually deforested [9]. The Poisson regression uses the following formula: where: E( | ) = the expected count of deforestation (Y) given certain input conditions ( ) = independent variable (X0=1 for the constant term) = variable coefficients (or parameters) The model parameters consist of external variables (economic variables) and parameters that affect the price of agricultural products. Net Present Value formula: …….. (2) where: Bt = total revenue generated in year t, Ct = total costs in year t, i = interest rate T = expected lifetime (5 years)

Deforestation
The Deforestation that occurred at mineral forests was higher than at peatland forests because people prefer to utilize forests in mineral land first, where accessibility is easier and the existence of forests is also wider. Reduced forests in mineral land would then trigger people to take advantage of peatland forests. The deforestation was relatively similar to the results of [13]. Refer to [13]  The rate of deforestation at Kalimantan and Sumatra Islands varied depending on the level of spatial resolution of data sources used. Research used Landsat Image data, therefore he got larger amount of deforestation. This was because spatial resolution of the image was 30 m, more meticulous than the global data used in this study with spatial resolution of 3 km [11]. The deforestation in 2005-2010 happened as a result of government policy in the development of agricultural areas, the development of oil palm plantations and industrial plantations. This is in line with the findings of study of expansion of agricultural policy, timber extraction and infrastructure expansion [9]. The main reasons of forest cover deficit in Kalimantan were related to the expansion of worldwide markets for pulp, wood and palm oil [15,17]. While Margono [12] asserted that in the period of 2000-2010 the cause of deforestation was the expansion of agricultural areas, especially palm oil plantations, expansion of pulp and paper plantation industrial areas and industrial forest clearance. Based on the figure 4 areas with relatively flat up to undulating topography and relatively easy accesibility (with existing rivers), it is a priority area for forest exploitation, thus causing the area to have higher deforestation rates (yellow to red). ://dx.doi.org/10.22161/ijeab/3.3.9  ISSN: 2456-1878 www.ijeab.com Fig. 4

: Map of Deforestation at Indonesia Country.
While areas with topographic hills to mountains (the existence of roads is very limited), then the area of forest is still relatively not yet logged, so rate of deforestation is relatively lower (blue to green).  -3, Issue-3, May-June-2018  http://dx.doi.org/10.22161/ijeab/3.3.9  ISSN: 2456-1878 www.ijeab.com

Page | 782
Based on figure 5, at Sumatra island, emission factors in Riau Province, Riau Islands, South Sumatra and Bangka Belitung have relatively higher value compared to other provinces. This is related to the existence of large peatland forest located in the area. Conversion of peatland forest into palm oil plantations causes the carbon emission factor to be higher. Meanwhile at Kalimantan island, carbon emission factors in West Kalimantan Province and Central Kalimantan have relatively higher value compared to other provinces. This is related to the presence of large peatland forest located in this area, while peatland is the highest contributor to emissions. Implementation The REDD policy at Indonesia, Sumatra island and Kalimantan island as presented in figure 6 and table 3. Fig. 6

International Journal of Environment, Agriculture and Biotechnology (IJEAB)
Vol -3, Issue-3, May-June-2018  http://dx.doi.org/10.22161/ijeab/3.3.9  ISSN: 2456-1878 www.ijeab.com Refer table 3, the forest emissions (emitable CO2) at Indonesia was 84.64 million tCO2e, donation from peat land forest 28.50 million tCO2e (33.37%) and from mineral forest 56.14 million tCO2e (66.33%). Based on spatial distribution, the forest emissions (emitable CO2) at Sumatra Island was 16.79 million tCO2e, donation from peatland forest 6.80 million tCO2e (40,48%) and from mineral forest 9.99 million tCO2e (59,52%). Meanwhile the forest emissions (emitable CO2) at Kalimantan Island was 27.16 million tCO2e, donation from peatland forest 8.76 million tCO2e (32.25%) and from mineral forest 18.40 million tCO2e (67.75%) Impact of REDD policy in Indonesia targeted carbon emissions of 4.41 million ha. Meanwhile, the gross emission reduction that could be obtained was 3.32 million tCO2e, and emission that could be absorbed by forests was 1.09 million tCO2e. Distribution on Sumatra island, targeted carbon emissions of 1.66 million ha. Meanwhile, the gross emission reduction that could be obtained was 0.88 million tCO2e, and emissions that could be absorbed by forests was 0.79 million tCO2e. Meanwhile implementation REDD policy at Kalimantan island, targeted carbon emissions of 1.10 million ha. The gross emission reduction that could be obtained was 0.85 million tCO2e, and emissions that could be absorbed by forests was 0.27 million tCO2e. Both islands (Kalimantan and Sumatra) contribute carbon emissions as much as 69.14%. Meanwhile, according to [1] stated that Indonesia had various emission levels from deforestation on each island. The highest emissions came from Sumatra, which were almost 56% of all emissions, and the second was Kalimantan with 28%, thus total for both islands was 84%. Therefore, it is important to focus on these two islands in implementing emission reduction strategies. The high emissions from Sumatra and Kalimantan were caused by the high deforestation rate on both islands, reaching 77% of Indonesia's total deforestation. Meanwhile [2] Deforestation in Sumatra contributed the greatest importance of the existing focus on clearance of peatland forest. The REDD policy was capable of reducing carbon emissions at Indonesia by 1.09 million tCO2e (24.753%). Meanwhile, the reduction of carbon emission in peatland forest area was 0.86 million tCO2e (28.30%) and in mineral soil forest area was 0.23 million tCO2e (14.02 %). The REDD policy was capable of reducing carbon emissions at Sumatra Island by 0.78 million tCO2e Results of the study [2] that calculated carbon emissions in Bolivia, GeOSIRIS could also be used to evaluate how much reduction of deforestation could be achieved with the price of alternative carbon. Refer [1] with international CO2 price of US$ 5-50 /tCO2, we can simulation relationship carbon price with deforestation and emission at the Kalimantan Island, Sumatra and Indonesia as show on figure 7. Based on figure 7, with a price of $10 it could be reduced by about 17 % -30% and at $50 by around 40 % -70%. The increase in carbon prices will spur activities to protect the forests so that the forests will be better protected and deforestation will also occur. Conversely, if there is an increase in price of agricultural products, then the rate of deforestation will also increase, because more forest areas will be cultivated into agricultural areas. The relationship between carbon prices to deforestation and carbon emissions has the same pattern (refer to fig 7). The impact of rising carbon prices leads to increased deforestation as well as carbon emissions. The impact of rising carbon prices on forest areas in Sumatra has a bigger impact than deforestation on the average of Indonesia and also forests in Kalimantan. Similarly, a success in reducing deforestation is linearly related to reduction of carbon emissions. The more forests that can be protected from logging, the more economically beneficial they will be a. Price with Deforestation b. Price with Carbon Emission Fig.7

IV.
CONCLUSION In the period 2005-2010, deforestation at Indonesia was 4.65 million ha (4.99 %). The simulation result, impact of REDD policy could reduce deforestation at Indonesia by 0.66 million ha (17.45%). With assumption that international carbon price of US$ 10/tCO2e, the change of emissions due to REDD was 24.75%, or reduced emissions by 1.09 million tCO2e/5 years. Finally, Gross National Revenue from carbon payments (NPV 5 years) was US$ 10.917 billion, where incentivize emission reductions to sub-national entities (NPV, 5 years) was US$ 9.178 billion and net central government surplus from carbon payments was US$ 1.739 billion (NPV, 5 years).