The estimation of above-ground oil palm carbon stocks of different ages through vegetation index

Above-ground carbon stocks play an important role to support the sustainability of oil palm plantation, and vary with soil characteristics, climate conditions, and agronomic practices. The common direct approach to assess aboveground carbon stocks may be time-consuming, and laborious, so may not be suitable especially for large scale observation. Therefore, the use of remote sensing may overcome some limitations coming up for conventional methods, yet requires a considerable data validation to obtain a reliable model. In this study, the normalized difference vegetation index (NDVI) based on Sentinel-2 satellite imagery, acquired in September 2023, was used in combination with field measurements, i.e. biomass estimation, investigated in three separated of oil palm plantations differing in ages, i.e. young (< 10 years), middle (10 – 20 years) and old (> 20 years), subjected with similar soil characteristics, and climate condition, in Bojong Datar, Cibungur, and Cikasungka, West Java. Results show that above-ground carbon stocks increase with oil palm ages, i.e. from young to middle ages, with value ranging from 3.40 – 65.70 ton/ha, yet decreases down to 17 ton/ha with ages > 20 years. Similarly, it also occurred for NDVI index, i.e. increase from 0.50 – 0.59 for young to middle of oil palms, and down to 0.51 for oil palm > 20 years old. This study suggests a strong correlation between carbon stocks and NDVI values (R2 = 0.878), pointing out the potential use of vegetation index to estimate above-ground oil palm carbon stocks. In the future, with further validation, the use of NDVI to estimate plant biomass may provide a potential substitute for the field measurement methods with a considerable higher efficiency and accuracy.


Introduction
Above-ground carbon stocks, derived from plant biomass, has received considerable attention to study carbon dynamics in the ecosystem, and demonstrate an important indicator to investigate the potential carbon sequestration [1], productivity and sustainability of land, and takes into account the largest part of above-ground carbon stocks [2].It provides a substantial contribution to the global carbon cycle through CO2 fixation and storage in the stems, roots, and soil, and subsequently released during the process of biomass decomposition when plants die or are burnt.Above-ground carbon stocks also provide an approach to estimate the potential below-ground components such as roots, rhizomes, and soil microbes [3].
As CO2 provides a considerable contribution to the greenhouse gases [4], so the observation of plant biomass related carbon stocks may be used, to some extent, to forecast the atmospheric CO2 for instance due to land conversion, and estimate the potential amount of CO2 sequestered into terrestrial land.Though the current potential CO2 emission from Agriculture and Forestry sectors, ca.10% from total atmospheric CO 2 concentration, yet their contribution is quite important, estimated around 25% to total terrestrial carbon stocks [5], and may be more sensitive to the change in the biophysical environment, such as global warming, in compare to different carbon pools.However, global warming phenomenon has raised to the concern on the need for high-quality monitoring systems to determine terrestrial carbon level, and its dynamic through time with corresponding to different physical environment conditions, and landuse.The use of current developed direct field measurement to estimate above-ground carbon stocks, i.e. destructive and non-destructive methods, exhibits some problems in terms of limited size of tree and area samples, time and resource consuming, and infeasible for a large scale analysis, though these methods are considered to provide more accurate estimates [6].From the two methods, the non-destructive approach such as allometric estimation [7], may be more preferable as it is less time consuming and less expensive, but requires some validation in terms of diameter and weight of tree components.
To deal with some field measurement limitations observed during the practical implementation towards smart and precision agriculture [1], remote sensing technology may provide some advantageous related to its efficiency in the context of a larger area coverage with corresponding to shorter operational time.For above-ground carbon stocks measurement, the tool does not estimate the absolute number immediately, yet it measures the parameters that linked to plant biomass such as tree height, stem diameter, leaf area index, crown size and so on.Therefore, the carbon stocks estimation is obtained from remotely sensed data validated with field measurement, and furthermore can be extrapolated to different areas with more limited data availability.
The remote sensing technique through vegetation index analysis has been reported to have a strong association with plant biomass, as indicator of above-ground carbon stocks [1,3,8].The most common vegetation index used in the vegetation monitoring such as normalized difference vegetation index (NDVI), with value ranging from -1 to 1.The index indicates the higher the NDVI value, the higher the vegetation density.In this research, NDVI technique is used to estimate above-ground carbon stock in oil palm plantation.

Study site
Three different oil palm plantations located in Bojong Datar, Cibungur, and Cikasungka, West Java were observed to reveal the above-ground carbon stocks (Fig 1).The study included field measurement to have a direct assessment of oil palm performance, including stem diameter, stem and tree height.Additionally, satellite image analysis is performed to evaluate NDVI value, and used to correlate the carbon stocks predicted from field measurement.

Environmental background
The average precipitation data (2018 -2022) demonstrate rainfall of up to 2600 mm per year in Bojong Datar, 2950 mm per year in Cikasungka, and 3300 mm per year in Cibungur.The intensity of rainfall is a bit lower in the first semester, mainly on June to August, and increase in, gradually, the second semester.With subjected to one dry month period (< 60 mm), the climate condition is a slightly drier in Bojong Datar in compare to different two locations.

Field measurement
The above-ground carbon stock was measured based on nine class of oil palm ages, i.e.Plant biomass with height > 1.30 m was calculated through formula as below [8]: (1) With: Y : plant biomass (kg dry weight/tree) D : diameter at breast height (cm) H : height of tree trunk (m) While for plant biomass with height < 1.30 m, the formula as below: (2) With: Y : plant biomass (kg dry weight/tree) H : height of tree top (m) Following plant biomass measurement, the above-ground carbon stock is measured using formula as below: (3) Cb : Carbon biomass (ton) B : total plant biomass (ton) OC : organic carbon (%), which is assumed 50% from plant biomass Prior to NDVI analysis, image pre-processing was done using Google Earth Engine (GEE), i.e. cloud masking, to make sure that the pixels in the image are transparent resulting from the reduced cloud cover as a noise.The NDVI is calculated based on reflectance value at band 8 (NIR) and band 4 (Red) using the following formula: With: NIR : near infrared reflectance Red : red reflectance NDVI : index value ranging from -1 to + 1

Analysis of statistics
To evaluate the correlation between NDVI and carbon stocks, the analysis is performed through linear regression y = a + bx, with y = carbon stock estimation, and x = NDVI value.To obtain both a and b value, below formula is used: While for a value can be determined through: Coefficient correlation (r) is calculated using the following formula: The closer r value to 1, the better correlation can be expected between the two variables.The output model will be applied for above-ground carbon stocks measurement through NDVI estimation.The carbon stock map is classified based on several range, i.e. < 15 ton C/ha, 15-25 ton C/ha, 26-70 ton C/ha, and >70 ton C/ha based on equal interval method.

Results and Discussion
In this study, the above-ground carbon stocks are estimated through plant biomass investigation.It has been widely used as an approach to estimate plant stored carbon stock [1,3,8].In this case, the dynamics and important role of carbondioxide (CO2) is revealed especially through process of photosynthesis, and metabolism leading to a considerable amount of carbon storage in the plant.

Above-ground carbon stock estimation based on different oil palm ages
The above-ground carbon stock average increased from 3.40 tons/ha to 65.70 tons/ha with correspond to the increase in terms of ages from 3 years to 14 years old of oil palm, then decreased down to 17 tons/ha at 27 years old of oil palm.In the similar physical environmental condition, a slight lower carbon content, with concomitant plant biomass, revealed on 18 years old of oil palm than the 19 years old may be attributable to the history of crop maintenance level such as fertilization, as well as pest and disease attacks, which is lower than the standard, leading to lower crop performance.The high carbon stock content of 14 years old of oil palm, which is linked to total biomass 131.40 ton/ha from 22.73 ha area, pose a net value of about 65.70 ton/ha.At 27 years old of oil palm, the amount of carbon stock estimated down to 17 tons/ha.Additionally, the lowest carbon stock obtained from 3 years old of oil palm with 3.40 tons carbon per ha.All these may explain the association of above-ground carbon stock more to oil palm ages than environmental variations, and provide a snapshot the dynamic balance between carbon fixation and carbon loss through respiration.

NDVI assessment based on different oil palm ages
The NDVI values demonstrate a non-linear relation to plant age, supposed to be influenced by canopy density and vegetation colour.To some extent, the NDVI values increase with the increase in the oil palm canopy density (Fig. 4).Area with younger oil palm population (< 9 years old) pose a NDVI values ranging from 0.50 -0.54, and it is increased to close to 0.60 to oil palm with ages 9 -19 years old.The decrease in the NDVI values, down to close to 0.51, occurred with the concomitant increase in the oil palm ages from 22 -27 years old.

Correlation results
Our analysis suggested the association between carbon biomass obtained from field assessment and the NDVI values through regression equation Y = 685.97x-336.92 with correlation coefficient (r)  However, 12% of factor may not be explained which is supposed to be different factor other than plant canopy density such as trunk and frond biomass which is not identified in this study.

4 .
Remote sensing data analysis Imagery from Sentinel 2, dated 2 September 2023, was used to analyze the vegetation index.Rainfall in the last month was ~ 0 mm, subjected to a prolonged dry period.The normalized difference vegetation index (NDVI) was used to measure the difference of two specific bands, i.e. near-infrared reflected by vegetation, and infrared which vegetation absorbs.

Figure 3 .
Figure 3. Field measurement of potential carbon stocks estimated from plant biomass of different ages of oil palm.

Figure 4 .
Figure 4. NDVI value of different ages of oil palm.
878.A high r value implied a strong correlation coefficient, and may be used to estimate the spatial distribution of carbon stocks in the study area.

Figure 5 .
Figure 5. Correlation between above-ground carbon stock and NDVI value.

Figure 7 .
Figure 7. Spatial distribution of above-ground carbon stocks derived from NDVI in oil palm plantation in Bojong Datar (left) and Cikasungka (right).
Above-ground carbon stocks demonstrate a strong correlation with ages with the highest potential of carbon storage identified in 14 years old of oil palm, i.e. 65.70 tons C/ha.The NDVI values for oil palm population ranging from 3 -27 years old are 0.50 -0.59, and suggested to be attributable to plant canopy density.The coefficient correlation of above-ground carbon stocks and NDVI 0.87 with unexplainable the remaining factors such as trunk and frond biomass, as well as physical environment.The most of above-ground carbon stocks stored in the range of 26 -70 tons/ha, and with strong influence from plant ages.Additional validation of model in different physical environmental conditions is urgently required to improve the accuracy.

Figure 6 .
Figure 6.Spatial distribution of above-ground carbon stocks derived from NDVI in oil palm plantation in Cibungur.

Table 1 .
Carbon stock and NDVI values through different oil palm ages.
C/ha.The high potential above-ground carbon stocks in Cikasungka oil palm plantation may be related to higher population of productive oil palm ranging from 11 -22 years old.Similarly, the most area in Bojong Datar oil palm plantation, ca.60%, and Cibungur oil palm plantation, ca.50%, contributed to 26 -70 tons C/ha with oil palm ages ranging from 3 -19 years old.