Biomass Stock Estimation Using Landsat 8 Imagery in Bukit Tigapuluh National Park, Riau

Forests certainly store biomass content which is reflected in the physical appearance of a tree. In calculating biomass directly, direct surveys and measurements are needed. Remote sensing technology, in this case, is a tool for monitoring and calculating the biomass content of vegetation. In estimating biomass using remote sensing, the biomass content of vegetation in the field is still needed. This study aimed to estimate the biomass content of the Meranti plant (Shorea parvifolia Dyer), which is the dominant plant species in the Bukit Tigapuluh National Park in Indragiri Hulu and Indragiri Hilir Regencies, Riau Province and Tebo Regencies and Tanjung Jabung Barat Regencies in Jambi Province with using Landsat 8 imagery. Field measurements and remote sensing images, which in this case are vegetation indices in the form of NDVI, ARVI, GNDVI, MSAVI2, and EVI, will produce a biomass estimation model, and the most suitable model will be selectively selected based on the strength of the relationship between the two parameters. The resulting model between the biomass value and the vegetation index will then become an estimate of the biomass of the dominant plant species in Bukit Tigapuluh National Park.


Introduction
The usage of fossil fuels (BBF) and changes in land use are particularly responsible for the global phenomenon of climate change [1].Various parties have widely discussed this phenomenon, and the effects of this phenomenon are due to the shrinking of tropical forests and industrial emission pollution to global environmental damage [2].Using trees inherent capacities as CO2 absorbers is one of the most efficient strategies to minimize greenhouse gas emissions [3].1291 (2024) 012019 IOP Publishing doi:10.1088/1755-1315/1291/1/012019 2 [4] state that to reduce the impact of climate change, Indonesia can increase carbon stocks in the Reducing Emission from Deforestation and Forest Degradation Plus (REDD+) scheme through sustainable forest management (SFM) by: 1. Lowering carbon emissions caused by logging (in forest conversion) 2. Lowering emissions related to forest degradation (in sustainable forest management practices) 3. Retaining emissions/carbon stocks (in conservation forests) 4. Increasing carbon stocks (in reforestation and ecosystem restoration activities) Approximately 90% of the biomass in forests on Earth's surface consists of wood, branches, leaves, roots, forest litter, animals, and microbes [5,6].The process of photosynthesis, in which carbon dioxide (CO2) in the atmosphere is bound and transformed into an energy source (sugar clusters) helpful for life, allows forests to absorb carbon dioxide from the environment.In the form of biomass, plants store this energy [7].The process of photosynthesis, which is aided by sunshine, soil water, and chlorophyll, allows vegetation to absorb carbon dioxide (CO2) from the atmosphere, plays a part in efforts to boost carbon dioxide absorption in forests [8].The negative effects of the greenhouse effect can be reduced the more CO2 plants absorb and store as carbon biomass [9].
The carbon stock estimation method that has been developed is based on field measurements at the plot level.Carbon stocks are estimated from biomass by following the rule that 40% of biomass is carbon [10].To estimate carbon stocks in a larger area, a way is needed to extrapolate the results of plot-based measurements to the landscape level.One potential method to meet these needs is to use remote sensing technology [6].One data from remote sensing is reflectance and spectral characteristics [11].Currently, many are being developed research to estimate carbon stocks using active remote sensing systems, namely by using radar-based satellite imagery.
Availability of time series image data covering all regions of Indonesia, free for download, and relatively good resolution (spatial, temporal, and radiometric) are three advantages owned simultaneously by Landsat 8 and not owned by other images so that it supports various types of images.Needs, one of which is in the study of vegetation.Landsat 8 satellite imagery, with all kinds of advantages, is used as remote sensing data to estimate carbon stocks [12].The vegetation index is an algorithm that is used to highlight aspects of plant density or other aspects linked to density, such as biomass, Leaf Area Index (LAI), chlorophyll concentration, and so on, in images (often multichannel images) [13].Further research is necessary to determine which vegetation index can be utilized to create the best model for predicting biomass in Bukit Tiga Puluh National Park Forest because there are numerous vegetation indices.The aims of this research are as assessing the relationship between the value of the vegetation index with biomass and forest carbon stocks from the field data and develop the best model for biomass estimation of Bukit Tiga Puluh National Park Forest based on the value of the vegetation index.

Research Area
Bukit Tiga Puluh National Park is a National Park Area that has been established based on the Decree of the Director General of Forest Protection and Nature Conversion, number SK.79 76/IV-KKBHL/2015 concerning the register number of nature reserve areas, nature conservation areas and hunting parks [14].The Bukit Tiga Pulih National Park area is in two provinces, namely Riau Province in Indragiri Hulu Regency and Indragiri Hilir Regency, and Jambi Province in Tebo Regency and West Tanjung Jabung Regency.The area is quite large with an area of 144,993 hectares [15].The area has a flat topography and is categorized as a tropical rainforest which has a climate that is always wet with rainfall of more than 2500 mm per year, the soil is quite dry and has an altitude ranging from 60-834 masl.

Data Collection
The biomass estimation carried out in this study focuses on Meranti species (Shorea parvifolia Dyer) which is one of the dominant plant species in Bukit Tiga Puluh National Park [16].Biomass estimation is carried out by utilizing data from Landsat 8 Collection 2 tier 1 Surface Reflectance imagery which has been corrected for the atmosphere.The proportion of incoming solar energy that is reflected from the earth's surface has been measured by Landsat 8 Collection 2 SR.Image acquisition was carried out in the 2016-2017 period using the median function to obtain images with low cloud cover presentation.Obtaining the Meranti species biomass value is carried out directly in the Bukit Barisan National Park by calculating using an Allometric equation that has been determined based on the observation location.

Data Processing 2.3.1. Vegetation Index Transformation
In the regression analysis using the field-measured biomass data, the transformation of the vegetation index value is used as an independent variable.There are five types of vegetation indices that are processed to obtain the best method for estimating biomass based on the selected vegetation index, namely NDVI, ARVI, GNDVI, MSAVI2 and EVI.The five indices have calculation methods with different values to obtain pixel values which will then be regressed with the biomass data from field measurements as shown in Table 1.Enhanced Vegetation Index (EVI) [20]

Biomass Calculation Stage
Data obtained from measurements in the field are then calculated for above-surface biomass where in the process of calculating sub-surface biomass from the sample it is ignored.Field measurements were carried out by in situ sampling without harvesting (non-destructive sampling).Sample data obtained from direct measurements in the Bukit Tigapuluh National Park were then calculated using an allometric equation to obtain the biomass value of the sample.
Standard allometric equations that have been widely published are frequently used, however because their coefficients vary depending on the location, using them can produce large errors when processing biomass estimation [21].In this research using the equation according to [22] which is suitable for tropical forest locations.
In the allometric equation according to [22] there are 2 calculation parameters, namely density of wood species and diameter of breast high (DBH) where the equation is adjusted to the conditions at the measurement location so that the accuracy obtained is quite good.This equation is based on the location of the Bukit Tigapuluh National Park which is humid throughout the year. ( Which Y is Total Biomass (kg); DBH is Diameter Breast High (cm); and is wood density (g/cm 3 ).

Correlation and Regression
The pixel values obtained from image processing using several index methods are then tested between the biomass data from field measurements and the pixel values from several methods.In order to make the best decisions, correlation testing seeks to ascertain the strength of the relationship and contribution of the independent factors to the dependent variable.Next, a value of the dependent variable based on the independent variable is estimated using regression analysis to determine the value of some variables depending on the values of other variables [23] The regression model generated for each vegetation index can be determined from the description of the shape of the regression line on the regression graph which has the value of R (correlation coefficient), R² (coefficient of determination, ANOVA test and T test.The best results based on the regression model will then be used as a formula in estimating biomass based on the vegetation index which has most effective regression model).Based on the table, we can know that from the ranking results based on R and R2 values, the vegetation indices that occupy the top three positions are ARVI and NDVI, while the bottom three are EVI.ARVI (Atmospherically Resistant Vegetation Index) is intended as a vegetation index that is applied to areas with atmospheric aerosols, so that ARVI can avoid atmospheric intervention, especially at high concentrations of particulate pollution [24].This certainly has something to do with the research area which is near the equatorial latitude with high sunlight intensity which results in high water vapor in the evaporation and evapotranspiration processes.In addition, these results are also in line with research in Tesso Nilo National Park [25] where the best regression model for estimating biomass comes from the ARVI vegetation index.NDVI in this case takes third place.NDVI, which in this case has a combination of red band and NIR band, has the advantage of obtaining information from chlorophyll or the greenness of the existing leaves without any atmospheric correction [24].However, in the research area where land cover is homogeneous, the performance of NDVI is not optimal, because NDVI is sensitive to separate vegetated and non-vegetated areas [24] .These results are also in line with similar research [26]with the level of accuracy of biomass modeling using NDVI is lower than ARVI.EVI occupies the lowest position where the correlation between the estimated value of the model and the actual data is very low.Due of this, EVI cannot be used to estimate biomass.After the regression results were obtained, validation tests were carried out from the three best models on the validation data outside of the data used.As for this validation test, various types of validation will be carried out by giving a score according to the validation test.The logarithmic regression model based on NDVI has the minimum standard error of estimation, according to the model accuracy test results, making it the regression model with the best accuracy in comparison to other models.Therefore, the best model to estimate biomass is to use NDVI with regression model y = 830.36ln(x)+ 159.73.

Calculation of Biomass Stock
Based on the best regression model obtained, referring to the 3 best models of vegetation index, consisting of ARVI with logarithmic regression method yielding a model y = 370.57ln(x)+ 154.08, it is known that the total amount of forest biomass in the study area is 44,135.79tons.While ARVI with linear method produces a model of y = 509.24x-334.46, it is known that the total amount of forest biomass is 45.954.79 tons.Meanwhile, NDVI with logarithmic regression method produces a model y = 830.36ln(x)+ 159.73.It is known that the total amount of forest biomass is 45,615.78Tons.From the three models, it was found that the largest biomass stock estimation results were obtained from the vegetation index modeling from ARVI with a linear method.Forest carbon stocks are basically difficult to quantify with certainty, especially in a large study area.This is because the carbon pocket is very dynamic in absorbing, storing, and releasing carbon.Carbon stocks can be calculated through the tree biomass approach.Calculation of carbon stock in this study uses the assumption according to [27] that half of the biomass is carbon content.

Conclusion
The vegetation index value and biomass calculated from field data have a positive relationship, meaning that the higher the vegetation index value, the more directionally the biomass will be.This research was done in Bukit Tigapuluh National Park using Landsat 8 SR imagery.However, the type of vegetation index utilized determines how strongly the vegetation index correlates with biomass.From the resulting model, the NDVI (Normalized Difference Vegetation Index) with a logarithmic type of regression model is the best model for estimating biomass.This is because the research area was conducted in Bukit Tigapuluh National Park which is a primary forest area with a combination of bands on the NDVI index according to the conditions of the area.

Figure 1 .
Figure 1.Map of Research Area

Figure 2 .
Figure 2. Map of Vegetation Index

Figure 3 .
Figure 3. Map of Biomass Estimation Best Model (Source: Personal Data)

Table 1 .
Vegetation Index Formula

Table 2 .
Result Of Model Type Regression Vegetation Index (Source: Personal Data)

Table 3 .
Result of Validation (Source: Personal Data)

Table 4 .
Result of Biomass Stock Estiamtion (Source: Personal Data) Kwatrina R T and Antoko B S 2007 Rasionalisasi zonasi Taman Nasional Bukit Tigapuluh: penerapan kriteria dan indikator zonasi serta tingkat sensitivitas ekologi Jurnal Penelitian Hutan dan Konservasi Alam 4 391-407 [17] Rouse J.~W.Jr, Haas R ~H., Schell J ~A. and Deering D ~W. 1974 Monitoring Vegetation Systems in the Great Plains with Erts NASA Special Publication vol 351 p 309 [18] Kaufman Y J and Tanre D 1992 Atmospherically resistant vegetation index (ARVI) for EOS-MODIS IEEE transactions on Geoscience and Remote Sensing 30 261-70 [19] Gitelson A A and Merzlyak M N 1998 Remote sensing of chlorophyll concentration in higher plant leaves Advances in Space Research 22 689-92 [20] Qi J, Chehbouni A, Huete A R, Kerr Y H and Sorooshian S 1994 A modified soil adjusted vegetation index Remote Sens Environ 48 119-26 [21] Heiskanen J 2006 Estimating aboveground tree biomass and leaf area index in a mountain birch forest using ASTER satellite data Int J Remote Sens 27 1135-58 [22] Chave J, Andalo C, Brown S, Cairns M A, Chambers J Q, Eamus D, Fölster H, Fromard F, Higuchi N and Kira T 2005 Tree allometry and improved estimation of carbon stocks and balance in tropical forests Oecologia 145 87-99 [23] Karmila D, Jauhari A and Kanti R 2020 Estimasi Nilai Cadangan Karbon Menggunakan Analisis NDVI (Normalized Difference Vegetation Index) di KHDTK Universitas Lambung Mangkurat Jurnal Sylva Scienteae 3 451-9 [24] Somvanshi S S and Kumari M 2020 Comparative analysis of different vegetation indices with respect to atmospheric particulate pollution using sentinel data Applied