Carbon Emissions Prediction from Land Change Model in Bandung Regency, Indonesia

Carbon emissions in Bandung Regency is potentially increasing and can harm future environment in the area. Our study aims to predict carbon emissions in Bandung Regency based on land changes model of the area. We utilized Cellular Automata Markov method which predicted land use in the years of 2027 and 2036. This land use prediction used overlay intersect method and were analysed using two scenarios: backward and forward looking. The backward looking scenario analysis combined the land used map of 2019 and two land use maps prediction of 2027 and 2036. Meanwhile, the forward looking scenario analysis combined the land used map of 2019 and the urban planning plan of Bandung Regency in the years of 2016 to 2036. Carbon emission were calculate using stock difference method for both scenarios. Afterwards, the carbon emission results were classified using quartile method. The result in the backward scenario showed that the carbon emission is dominated compare to sequestration, with total emission of 811,330.81 tonnes of CO2 eq for the years of 2019 to 2027 and 998,288.22 tonnes of CO2 eq for the years of 2019 to 2036. In the forward looking scenario, carbon sequestration is dominated compare to emission, with highest sequestration around -804,125.22 tonnes of CO2 eq, while highest emission is around 245,506.51 tonnes CO2 eq. Overall, both simulation showed that the highest emission occurs in Cilengkrang-Cimenyan development area, and the carbon sequestration occurs in Baleendah, Cileunyi-Rancaekek, and Margahayu-Margaasih development areas.


Introduction
Global warming has been correlated with the occurrence of natural disasters in several areas in Indonesia, such as floods, landslides, droughts, and several ecological disasters that harm human livelihoods [1].It has been assumed, that anthropogenic activities are one of the contributor for the increase of global warming through high release of greenhouse gases into the atmosphere and reduced CO2 absorption by green areas.CO2 gas is the largest contributor based on data released by the Intergovernmental Panel on Climate Change (IPCC) and accounts for about 76 percent of total greenhouse gas emissions [2][3].Balanced amount of CO2 in the earth's atmosphere is needed to provide a warm effect on earth.However, if the amount is excessive, it will result in an imbalance between the energy received and released.This will affect the carbon cycle and has the potential to cause temperature 1245 (2023) 012004 IOP Publishing doi:10.1088/1755-1315/1245/1/012004 2 changes [4].Human activities that cause land changes can resulted in the reduction of green areas that serve as CO2 absorber which can then further increase the temperature in the area [5].The effects of global warming can be felt in local areas, especially in urban areas with high population density and activity.One of them is in Bandung Regency.This Regency observes annual population increase.For example, in 2016 the population of Bandung Regency was 3,596,623 people [6]; and increased to 3,666,156 people in 2021 with an area of 1,762.40km 2 .So that in 5 years there was an increase of 39.45 people/km 2 [7].The high population growth in Bandung Regency has resulted in an increasing need for housing which further trigger land changes.Green areas such as agriculture and plantations are often converted into residential areas.Continuity of this land change will have the potential for further increase of carbon emission in Bandung Regency and might cause negative impact on the environment, so proper mitigation to reduce carbon emissions is crucial.One way to identify proper mitigation approaches for reducing carbon emission in Bandung Regency is by predicting carbon emissions based on land change.This study aims to predict carbon emissions based on a spatial model of land change in Bandung Regency using backward and forward looking approaches.The model is dynamic in time so it can simulate or predict future conditions [8].Using this approach, we will be able to see the future trend of increasing emission values and identify areas that needs to be prioritized in carbon emission reduction actions.

Study Area
The location of this research is in Bandung Regency which is located at coordinates 107 0 22'-108 0 50' East Longitude and 60 0 41'-70 0 19' South Latitude.The total area of Bandung Regency is 1,762.4km 2 consisting of 31 districts.

Methodology
We performed two scenarios in this study: backward looking scenario and forward looking scenario, to see possible future carbon estimates.

Figure 2. Methodology
The steps are: land use prediction, land change estimation, carbon emission calculation, and analysis of carbon emission estimation based on land change.

Land use prediction.
Prediction of land use is carried out using data from the year 2019; and the time lapse data between the prediction year and the base years: 2004 and 2011.The data processing used Idrisi Selva software with the Markov Chain method, Cellular Automata Markov, and validation tests.Markov Chain analysis a pair of land maps in two-time series and produces a transition probability matrix, a transition area matrix, and a transition suitable image.The transition probability matrix is a matrix that records the probability that each land class will change to every other class.The transition area matrix is a matrix that records the number of pixels that are expected to change from each class during a specified time [9].The next process is to run the Cellular Automata Markov method to get land use predictions in 2027 and 2036 with input data for land use based on the year and two data from the previous Markov Chain.After obtaining the model, a validation test was carried out between the existing land use in 2019 and the results of the 2019 prediction model to find out how accurate the predictions were.

Land change estimation.
Land changes were processed by the intersect overlay method using ArcGIS software.Data processing is carried out in two scenarios by overlapping two land use map data.The backward looking scenario combines the land used map of 2019 land use with the predicted years 2027 and 2036; while the forward looking scenario combines the land used map of 2019 land use with the urban planning plan of Bandung Regency in the years of 2016 to 2036.After overlapping the two scenarios, the attribute data for land change is generated.

Carbon emission calculation.
There are several factors in the calculation of carbon emissions, some of the previous studies carried out: calculating carbon emissions from Municipal Solid Waste (MSW) using the method recommended by IPCC (2006) and divided into four parts according to the four treatment modes (sanitation landfill, simple landfill) [10]; Calculation of carbon emissions using the Logarithmic Mean Division Index (LMDI) model is used to describe the main factors which include the proportion of direct energy, the unit value of energy consumption, value creation effects, indirect carbon intensity, and output scale effects [11]; and identification of traces of carbon emissions based on household domestic use results using primary and secondary carbon emission calculations each year and the calculation of the Specific Emission Factor (SEF) for each household [12].This study, we calculated carbon emissions from land use change based on the Technical Guidelines Calculation of Baseline Emissions and Removals of Greenhouse Gases in the Land-Based Sector: Book I Scientific Basis [13], emissions are calculated following the IPCC Guidelines for National Greenhouse Gas Inventories 2006.Land that has changed over time is calculated based on the difference in carbon stocks between the initial year and the following years.Two approaches can be used: (1) Stock difference and (2) Gain and loss [14].

Analysis of carbon emission estimation based on land change.
To calculate the carbon emissions, we used the stock difference method for both scenarios.The stock difference is a calculation method by estimating the difference in carbon stocks at certain time intervals.Land use that does not change within a certain period of time consider to have no emissions or zero emissions, while land that has changed will emit a certain amount of carbon.The application of the calculation of carbon emissions are shown in Figure 3 where we multiply activity data in the form of area of land change and emission factors in the form of carbon stock values [14].The emission factor for each class uses national [13], sub-national emission factors [15], previous research [16], and assumptions that match the characteristics of other classes.The result of the calculation of carbon emissions based on land change is the result of the reduction between the initial land change carbon emissions and the next year's land change carbon emissions.

Source: IPCC (2006)
The analysis is based on how much carbon emissions are released for each land change.So that carbon emissions on the initial land use are reduced by carbon emissions from next year's land use.So, it can be concluded that if the reduction results in a negative emission number (-), sequestration will occur; positive (+) then emission occurs; and (0) which means non emission.Then to get the classification of carbon emission levels from the results of previous calculations using the quartile method.Classification using the quartile method uses the equation as shown in Equation (1).Classification of carbon emission levels will produce five levels: very low, low, medium, high, and very high.
Description: Qi : the value of i quartile; i : quartile type; and n : amount of data

Result and Discussion
The predictions of land use in 2027 and 2036 were based on simulations based on land changes occurred in 2011 and 2019.The prediction results for 2027 and 2036 can be used if the validation results meet the requirements.Validation was carried out between the existing year 2019 and the predicted year 2019 using the Cellular Automata-Markov method which were applied for predictions in 2027 and 2036.Prediction 2019 (Figure 4) is a simulation of land change in 2004 and 2011.The validation process in Idrisi Selva software obtained a Kappa value (Kstandard) of 0.87 or 87% which met the requirements greater than 70% [17].This value is included in the standard agreement percentage range of >75% which means the model is classified to be "very good" [18].Thus, the results of land use predictions in 2027 (Figure 5) and 2036 (Figure 6) can be used for the next process of land changes that occurred.

Total carbon emissions in backward looking scenario
For the backward scenario, land change is calculated from 2019 to 2027 and 2019 to 2036.Bandung Regency experienced changes in land use, either increase of land area, or land area decrease between 2019 to the predicted years of 2027 and 2036.Land use classification is divided into 14 classes, namely water (BA), airport (BDR), public facilities (FS), forest (HT), industrial and commercial (IP), mixed gardens (KC), moor (LD), bare land (LT), grassland (PR), plantation (KB), residential (PM), rice field (SW), shrubland (SB), and mining (TB).Land change from 2019 to 2027 (Table 1) emitted 811,330.81tonnes of CO2 eq (Table 2).The largest carbon emissions (marked in yellow in Tables 1 and 2) occurred due to the conversion of forest land to moor area of 4,157.01ha with emissions of 312,815.00tonnes of CO2 eq .Meanwhile, the largest sequestration (marked with green highlights in Tables 1 and 2) was caused by the change of rice fields into residential covering an area of 7,862.82ha with a sequestration value of -15,725.63 tonnes of CO2 eq .Table 1.Area of land change in Bandung Regency of 2019-2027 (ha) Table 2. Total carbon emissions in 2027 (ton CO2 eq ) Land change from 2019 to 2036 (Table 3) emitted 998,288.22 tonnes of CO2 eq (Table 4).The largest carbon emissions (marked in yellow in Tables 3 and 4) occurred due to the conversion of forest land to moor area of 4,618.67 ha with emissions of 347,555.19 tonnes of CO2 eq .Meanwhile, the largest sequestration (marked with green highlights in Tables 3 and 4

Total Carbon Emissions in Forward Looking Scenario
Land changes calculations comparing 2019 prediction map to the urban planning plan of Bandung Regency in the years of 2016 to 2036 (Table 5) tend to result in sequestration compared to emissions, as much as -804,125.22tonnes of CO2 eq (Table 6).The largest sequestration (marked by highlights in Table 5 and Table 6) occurred due to land conversion from shrubland to protected forest covering an area of 7,927.10ha with a sequestration of -334,048.16tonnes of CO 2 eq .The shrub is in a protected zone of the urban planning plan in Bandung regency.In spatial planning, land conversion in protected zones is limited.The shrub land in the protected zones probably originated from forest.Their fuction had changed due to illegal logging activities.The absence of reforestation efforts has resulted overgrown shrubs in the former logging area, and it takes a long time within natural process to return into forest.
Based on the spatial policy, shrubs that are inside the protected zone will be reallocated to forest [19].
Based on Government Regulation of The Republic of Indonesia Number 24 (2010), it is about the use of forest area.It explains that the reforestation is an effort to plant type of forest trees species in the areas where forests are ruin.Hence, it restores their function again.These areas are bare land, reeds, or shrubs [20].Meanwhile, the largest emission (marked with yellow highlights in Table 5 and Table 6) occurred due to the conversion of forest land to protected forest covering an area of 18,726.66ha with emissions of 245,506.51tonnes of CO2 eq .We use a regulation from the Ministry of Environment which states that the emission factor for forest land cover types on a regional scale (Java Island) has higher secondary forest carbon stocks than primary forest carbon stocks [15].The classification of carbon emission levels is detailed in Table 7.The carbon emission classification map can be seen in Figure 7, Figure 8, and Figure 9.The largest total carbon emissions and very high levels in the backward looking scenario occurs in forest land that was converted to moorlands because forests have a higher carbon stock value than moors (85.25 t/ha and 10 t/ha, respectively).This indicates stock decrease which will lead to further emissions.The forward looking scenario tends to result in sequestration due to land changes from low carbon stocks to high carbon stocks, with the largest sequestration occurs on shrubland with a carbon stock value of 30 t/ha to protected forest with a carbon stock value of 72.14 t/ha.The Cilengkrang-Cimenyan area is the largest emitter in both backward and forward-looking scenarios.This is because that forest land, moors, and mixed gardens have high carbon stocks have been turned into residential areas.Meanwhile, the Baleendah, Cileunyi-Rancaekek, and Margahayu-Margaasih areas produced sequestration compared to emissions in both backward and forward-looking scenarios.This is because domination of changes rice fields to residential areas.The residential carbon stocks are higher compare to rice fields.The residential carbon stock value is 4 t/ha, and the rice fields value is 2 t/ha [13].The higher carbon stock value could relate to the property developer's obligation to provide green areas of 30 percent of the total development area.It ensures the ecosystem balance, and the carbon cycle in the residential environment must be good [21].Meanwhile, rice fields not only have a low carbon stock but are also considered as one of the main sources of methane emissions.However, regarding the carbon stock of residential and rice fields, further studies need to be done.Overall, we found that the forward looking scenario is a good approach to predict carbon emissions resulting from land changes that occurred in the backward looking scenario.We can also conclude that the spatial pattern plan that has been prepared in the urban planning plan of Bandung Regency in the years of 2016 to 2036 is a spatial planning policy that promotes environmental conservation through low emission areas.The results of this study can also be used as an input for stakeholders in formulating an action plan to reduce carbon emissions in the strategic areas, such as in the Cilengkrang-Cimenyan area as the largest emission contributor area.The Bandung Regency Government is also advised to monitor and act decisively on irregularities in land change and protect agricultural land.

Conclusion
We found that up to the backward looking scenario showed emission of 811,330.81tonnes of CO2 eq from land change in 2019 to 2027, and emissions of 998288.22tonnes of CO2 eq from land change in 2019 to 2036.Forward looking scenarios tend to result in sequestration compared to emissions of -804,125.22tonnes of CO2 eq with the largest emission 245,506.51tonnes of CO2 eq .The Cilengkrang-Cimenyan area is the largest contributor to emissions because forest land, moor, and mixed gardens that have high carbon stocks are converted into residential, while the Baleendah, Cileunyi-Rancaekek, and Margahayu-Margaasih areas are the sequestration contributors because they are dominated by conversion of rice fields into residential areas.Forward looking scenarios can anticipate predictions of carbon emissions due to land changes that occurred in backward looking scenarios.

Figure 1 .
Figure 1.Study area in Bandung Regency 2.2.Data The data used in this study are 1) three-time series of land use in Bandung Regency in 2004, 2011, and 2019; 2) the urban planning plan of Bandung Regency in the years of 2016 to 2036.Both data are in 1:50.000scale shapefile format obtained from the Public Works and Spatial Planning Office.

Figure 7 .
Figure 7. Carbon emission map 2019 -the urban planning plan

Table 3 .
) was caused by the change of rice fields into residential areas covering 11,146.81hawith a sequestration value of -22,293.61tonnes of CO2 eq .Area of land change in Bandung Regency of2019-2036 (ha)

Table 4 .
Total carbon emissions in 2036 (ton of CO2 eq )

Table 5 .
Area of land change in Bandung Regency of 2019 -urban planning plan (ha)

Table 6 .
Total carbon emissions in urban planning plan (ton of CO2 eq )

Table 7 .
Carbon emission level classification