Keywords

1 Introduction

Carbon which is an important element required by all the living things on earth is stored in different places and in different forms. Thus, the amount of carbon stored in a system is known as “Carbon Stock” of Carbon Pool”. Carbon sequestration is a phenomenon for the storage of CO2 or other forms of carbon to mitigate global warming. There are major carbon pools i.e., ocean, soil, atmosphere, and forests. In forest ecosystem carbon is stored as aboveground biomass (leaves, trunks, limbs), belowground biomass (roots), deadwood, litter (fallen leaves, stems), and soils. For mitigation of climate change and global warming, sustainable management of carbon pools is important [1,2,3]. Various human activities like afforestation and soil degradation constantly cause reductions in carbon stock and its sequestration. Thus, several researchers studied carbon stock and its sequestration through forest [4, 5] and soils [6, 7]. Ray et al. [8] did a study on estimation of carbon stock and sequestration at the mangrove forest of Sundarbans in the Bay of Bengal, India. They observed that carbon stock is lower in the tropical mangrove forest than in the terrestrial tropical forest and their annual increase exhibits faster turnover than the tropical forest. Feng et al. [9] studied soil carbon stock and its sequestration in China. They conclude that accumulation of soil carbon stock will not necessarily increase the amount of decomposition in warm climate; however, it will increase the productivity of crop land and its ecosystem functions. The integrated approach of GIS (Geographical Information System) techniques with satellite data is highly potential to generate reliable information on natural ground surface data i.e., land surface temperature, LULC (landcover/landuse), NDVI (normalized difference vegetation index). Using this generated information carbon stock [10] and its sequestration [11] can be calculated and is used for its sustainable management [12,13,14]. Bordoloi et al. [14] conducted a study to model the carbon stock and its sequestration using integrated approach of GIS techniques with satellite data in North Eastern part of India. They have shown that the use of different optical satellite derived vegetation index i.e., NDVI, SAVI (soil adjusted vegetation index), ARVI (atmospherically resistant vegetation index), and empirical modeling approach in the study is effective. In this present study, estimation of carbon stock and its sequestration is conducted in Imphal west district of Manipur state using integrated approach of GIS techniques with satellite data and open source tool InVEST model v3.5.0 (Integrated Valuation of Ecosystem Services and Trade-offs) for different types of LULC (landuse/landcover).

2 Study Area

The location of study area (Imphal West district) is provided in Fig. 1. The area of the district measured 558 Km2 and it lies at a latitude of 24.30–25.00N and longitude of 93.45–94.15E. Imphal West has the highest population of 2,21,422 among the other district of the state Manipur. And most of the time, it enjoys the comfortable weather.

Fig. 1
A set of 3 maps. The topmost map is of India. An arrow from the northeastern part points to the next map, which is of Manipur. Another arrow from the heart of Manipur points to the bottom map of Imphal West, located near 94 degrees east longitude and below 25 degrees 10 minutes North latitude.

Study area

Full size image

3 Data and Tools

The data used with its source of collection and extraction is provided in Table 1. Soil map prepared by National Bureau of Soil Survey & Land Use Planning (NBSS & LUP) on 1:500,000 scale was used for extraction of physical and chemical properties of soil viz. soil depth, soil texture, soil drainage, and soil erosion. Default carbon value is prepared by Intergovernmental panel for climate change based upon that holding capacity in different pools like Above ground biomass, Below ground biomass, Soil carbon, Dead wood, and Harvest wood products. LULC is generated from Landsat 5 TM and Landsat 8 OLI for the years 1996 and 2016 respectively by maximum likelihood supervised classification with the help of ArcGIS® software.

Table 1 Data used in the study
Full size table

The elevation, slope and aspect map for the study area were generated from SRTM-DEM (shuttle radar topography mission—digital elevation model) (30 m resolution). Prediction of future LULC for the year 2026 was achieved through GeoSOS-FLUS (geographical simulation and optimization system—future land use simulation) which consists of two main parts as ANN-based (artificial neural networks) probability-of occurrence estimation module; and self-adaptive inertia and competition mechanism CA (cellular automata) module. In this study, ANN-based probability of occurrence estimation module was adopted. The input parameter used is aspect, elevation, slope, euclidian distance to Road, euclidian distance to settlement, and existing LULC (historical) map. Aspect and Slope for the district are obtained from DEM imagery which is taken from SRTM satellite. GeoSOS- FLUS software consists of two main parts. By using LULC and the default carbon value map carbon stock and carbon sequestration map were obtained.

4 Equations

4.1 Land Surface Temperature

It is calculated as:

$$\begin{aligned} & {\text{Land}}\;{\text{surface}}\;{\text{temperature}} \\ &\quad = \frac{{{\text{Brightness}}\;{\text{temperature}}}}{{\left[ {1 + \left( {\frac{{0.00115 \times {\text{Brightness}}\;{\text{temperature}}}}{0.4388}} \right) \times {\text{ln}}({\text{land}}\;{\text{surface}}\;{\text{temperature}})} \right]}} \end{aligned}$$
(1)
$${\text{Brighness temperature = }}\frac{{{\text{Y}}_{2} }}{{{\text{ln}}\left[ {\left( {\frac{{{\text{Y}}_{1} }}{{{\text{TOA}}}}} \right){ + 1}} \right]}} \, - { 273}{\text{.15}}$$
(2)

where, Y1 & Y2 are the thermal conversion constant and TOA is the Top of Atmospheric spectral radiance and is calculated as:

$${\text{TOA = X}}_{\text{L}} \times {\text{ P}}_{{\text{cal}}} {\text{ + R}}_{\text{L}}$$
(3)

where, XL is the band-specific multiplicative rescaling factor, Pcal  corresponds to band 10, RL is the band-specific additive rescaling factor.

Then, land surface emissivity is calculated as:

$${\text{Land}}\;{\text{surface}}\;{\text{emissitivity}} = 0.004 \times {\text{Proportion}}\;{\text{to}}\;{\text{vegetation}} + 0.986$$
(4)

Proportion of vegetation is calculated as:

$${\text{Propotion of vegetation = }}\left( {\frac{{{\text{NDVI}} - {\text{NDVIs}}}}{{{\text{NDVIv}} - {\text{NDVIs}}}}} \right)^{2}$$
(5)
$${\text{NDVI is given as:}}\;\;{\text{NDVI = }}\frac{{\left( {{\text{NIR}} - {\text{RED}}} \right)}}{{\left( {{\text{NIR}} + {\text{RED}}} \right)}}.$$
(6)

4.2 Accuracy Assessment for LULC

Accuracy of the classified feature is measured by the overall accuracy (OA) and Kappa coefficient (K) and is calculated as:

$${\text{OA = }}\left( {\frac{{\text{Summation of all Total number Correctly Classified Samples}}}{{\text{Total Numbers of Samples}}}} \right) \times 100\%$$
(7)
$$\begin{gathered} {\text{K}} = \frac{{\left( {{\text{Observed Accuracy }} - {\text{ Chance Agreement}}} \right)}}{{\left( {{1 } - {\text{ Chance Agreement}}} \right)}} \, \hfill \\ {\text{If K }} < \, 0.{\text{4 poor}}, \, 0.{4 } < {\text{ K }} < \, 0.{\text{75 good}},{\text{ K }} > \, 0.{\text{75 excellent}}{.} \hfill \\ \end{gathered}$$
(8)

4.3 Carbon Stock and Sequestration Model

The default values of carbon values (Intergovernmental panel for climate change, IPPC) for different carbon pools are provided in Table 2 for different LULC types.

Table 2 Default carbon value (IPPC)
Full size table

The carbon stock \({\text{CS}}_{x(i,j)}\) in a grid cell (i, j) is estimated as:

$${\text{CS}}_{x(i,j)} = {\text{A}}\left[ \begin{gathered} \left( {{\text{CS}}\_{\text{above}}} \right){\text{CS}}_{x(i,j)} + \left( {{\text{CS}}\_{\text{below}}} \right){\text{CS}}_{x(i,j)} \hfill \\ + \left( {{\text{CS}}\_{\text{soil}}} \right){\text{CS}}_{x(i,j)} + \left( {{\text{CS}}\_{\text{dead}}} \right){\text{CS}}_{x(i,j)} \hfill \\ \end{gathered} \right]$$
(9)

where, A is the area of the cell; C_above, C_below, C_soil, and C_dead are the above ground carbon, below ground carbon density, soil organic carbon density, and dead organic matter carbon stock. Total carbon stock (CS) and its sequestration (SS) are then estimated as:

$${\text{CS = }}\sum_{x = 1}^n {{\text{CS}}_{x(i,j)} }$$
(10)
$${\text{SS}}\;{ = }\;{\text{CS}}^{p_{2} } - {\text{CS}}^{p_{1} }$$
(11)

where, \({\text{C}}^{p_{2} }\) and \({\text{C}}^{p_1 }\) are the carbon stocks of p2 and p1 year respectively.

5 Results and Discussion

5.1 NDVI

It is noted (Fig. 2) that in the year 1996 to year 2016, there is a change in vegetation which shows a decrease in the vegetation cover during the last decade and its responses to climatic parameters. In the year 1996, the vegetation dynamics ranges from −0.1561 to 0.6521 and the density is decreasing in the year 2016 ranging from −0.3636 to 0.6418. It shows due to climatic parameter change vegetation dynamics also respond.

Fig. 2
A set of 2 maps of Imphal West. The left map represents vegetation density in 1996. The legend reads high 0.06522 and low minus 0.1562. The right map represents vegetation density in 2016. The legend reads high 0.6418 and low minus 0.3636. Low-density regions are more prominent on the right map.

NDVI of Imphal west district in year 1996 (left) and 2016 (right)

Full size image

5.2 Proportion of Vegetation

In the year 1996 to year 2016 there is a change in vegetation and it decreases in the vegetation cover during the last decade and its responses to climatic parameters (Fig. 3). In the year 1996 the proportion of vegetation ranges from 0.0639 to 0.5091 and the proportion is decreasing in the year 2016 ranging from 0.05919 to 0.4714. This will affect surface emissivity. This enormous variation will affect the carbon holding capacity of the vegetation and it lead the chance to global warming and climate change.

Fig. 3
A set of 2 maps of Imphal West. The left map represents the proportion of vegetation in 1996. The legend reads high 0.5092 and low minus 0.0639. The right map represents the proportion of vegetation in 2016. The legend reads high 0.4714 and low minus 0.0592.

Proportion of vegetation in the years 1996 (left) and 2016 (right)

Full size image

5.3 Land Surface Emissivity

Figure 4 shows the difference in emissivity during the year 1996 (Fig. 4 left) and year 2016 (Fig. 4 right).

Fig. 4
A set of 2 land surface emissivity maps of Imphal West. The left map represents 1996 trends. The legend reads high 0.9879 and low 0.9862. The right map represents 2016 trends. The legend reads high 0.4714 and low 0.0592. Low emissivity regions are more prominent on the right map.

Land surface emissivity in the years 1996 (left) and 2016 (right)

Full size image

5.4 Land Surface Temperature

Land surface temperature for the year 1996 and year 2016 is shown in Fig. 5. Here, it shows that the temperature variation depends on the vegetation dynamics. The LST is high in urban places with 20.4 °C and low at forest areas with 9.93 °C in the year 1996 and 22.1 and 10.7 °C in the year 2016 respectively.

Fig. 5
A set of 2 land surface temperature maps of Imphal West. The left map represents 1996 trends. The legend reads high 20.464 and low 9.935. The right map represents 2016 trends. The legend reads high 22.101 and low 10.730. High-temperature regions are more prominent on the right map.

Land surface temperature in the years 1996 (left) and 2016 (right)

Full size image

5.5 LULC

LULC for the years 1996 and 2016 are shown in Fig. 6 and is classified as dense forest, sparse forest, scrub/grass, crop land, built-up, and water bodies. The predicted LULC for the year 2026 is shown in Fig. 6 (bottom). In the accuracy assessment, the overall accuracy is obtained as (41 + 40 + 39 + 47 + 28 + 33)/282 * 100 = 79.25% and the corresponding Kappa coefficient as 0.78 which is in the acceptable range.

Fig. 6
A set of 3 land use land cover maps of Imphal West. The left, middle, and right maps represent 1996 trends, 2016 trends, and 2026 trends respectively. The legend reads sparse forest, dense forest, waterbodies, built up, scrub, and cropland. An increase in built-up is evident from 1996 to 2026.

LULC for 1996 (left), 2016 (right) and 2026 (bottom)

Full size image

The distribution of LULC (Km2) is compared for the years 1996, 2016 and 2026 and is shown in Table 3. It is evident that in the study region crop land (26–35%) and built area (17–30%) are highest and lowest by dense forest from the year 1996–2026. Built-up area increased to 12.9% and crop land decreases at 8.7% from 1996 to 2026 in the study region.

Table 3 Distribution of LULC in 1996, 2016 and 2026
Full size table

5.6 Estimated Carbon Stock and Sequestration

The amount of carbon stocks of the years 1996 (Fig. 7 left), 2016 (Fig. 7 right) and 2026 (Fig. 7 bottom). It ranges from 0 to maximum 26.6 mg of C for the year 1996 and 23.13 mg of C for the year 2016 and 21.27 mg of C for the year 2026 respectively.

Fig. 7
A set of 3 carbon stock maps of Imphal West. The left, middle, and right maps represent 1996 trends, 2016 trends, and 2026 trends, respectively. The legend for 1996 reads high 26.599 and low 0. The legend for 2016 reads high 23.13 and low 0. The legend for 2026 reads 21.279 and low 0.

Estimated carbon stock for 1996 (left), 2016 (right) and 2026 (bottom)

Full size image

Table 4 representing the amount of carbon stock for each class for different years. As shown in Table 4 for the year 1996 carbon stock has been calculated as dense forest (11.55%), sparse forest (12.04%), crop land (40.07%), scrub/grass (31.01%), built-up (5.33%), and water bodies (0%). For the year 2016 it is as dense forest (12.4%), sparse forest (12.19%), crop land (37.65%), scrub/grass (28.44%), built-up (9.32%), and water bodies (0%) while the total of carbon stock is 4086645 Mg of C. In 2026 and percentage of carbon stock is as dense forest (12.36%), sparse forest (13.97%), crop land (33.95%), scrub/grass (30.01%), built-up (9.71%), and water bodies (0%) and the total is 3717033 Mg of C. Carbon stocking capacity is changing due to different parameters and Fig. 8 (right) shows comparison carbon stock value in different period while Fig. 7 (right) shows the decreasing trend from 1996 to 2026 (19.72%). Predicted carbon sequestration for the year 2026 is presented in Fig. 9. The southern part of the district has greater potential carbon sequestration than the northern part; however, this region as a whole has low carbon sequestration capacity (range 0 to −1.850).

Table 4 Estimated total carbon stock (Tg of C) for different LULC of 1996, 2016 and 2026
Full size table
Fig. 8
A set of 2 graphs. The left graph is a triple-bar graph that plots total carbon stock versus land cover for 1996, 2016, and 2026. The right one is a line graph that plots the total carbon stock versus year. Croplands have maximum carbon stock. The carbon stock values decrease with the years.

Comparison of carbon stock value (Tg) for different years (left) and changes in carbon stock from 1996 to 2026 (right)

Full size image
Fig. 9
A carbon sequestration map of Imphal West in 2026. The legend reads high 0 and low minus 1.850. High and medium sequestration zones are more widespread, especially southwards from the center.

Predicted carbon sequestration for the year 2026

Full size image

6 Conclusions

The study region has poor potential for carbon stock as well as its sequestration as this region is rapidly urbanizing (built-up area increases at 13% approximately from 1996 to 2026). Thus, the amount of carbon stock is decreasing at 19.72% from 1996 to 2026. As forest area have an enormous change in carbon holding capacity, conservation or afforestation of forest area will help in maintaining the carbon stock in the region.