Korean Journal of Remote Sensing (대한원격탐사학회지)

Korean Society of Remote Sensing (대한원격탐사학회)
권호 표지
DOI QR코드

DOI QR Code

Actions to Expand the Use of Geospatial Data and Satellite Imagery for Improved Estimation of Carbon Sinks in the LULUCF Sector

  • Ji-Ae Jung (Division for Environmental Planning, Water and Land Research Group, Korea Environment Institute) ;
  • Yoonrang Cho (Center for Environmental Data Strategy, Planning & Coordination Office, Korea Environment Institute) ;
  • Sunmin Lee (Center for Environmental Assessment Monitoring, Environmental Assessment Group, Korea Environment Institute) ;
  • Moung-Jin Lee (Division for Environmental Planning, Water and Land Research Group, Korea Environment Institute)
  • Received : 2024.04.08
  • Accepted : 2024.04.23
  • Published : 2024.04.30
Download PDF
⟨ Previous Next ⟩

Abstract

The Land Use, Land-Use Change and Forestry (LULUCF) sector of the National Greenhouse Gas Inventory is crucial for obtaining data on carbon sinks, necessitating accurate estimations. This study analyzes cases of countries applying the LULUCF sector at the Tier 3 level to propose enhanced methodologies for carbon sink estimation. In nations like Japan and Western Europe, satellite spatial information such as SPOT, Landsat, and Light Detection and Ranging (LiDAR)is used alongside national statistical data to estimate LULUCF. However, in Korea, the lack of land use change data and the absence of integrated management by category, measurement is predominantly conducted at the Tier 1 level, except for certain forest areas. In this study, Space-borne LiDAR Global Ecosystem Dynamics Investigation (GEDI) was used to calculate forest canopy heights based on Relative Height 100 (RH100) in the cities of Icheon, Gwangju, and Yeoju in Gyeonggi Province, Korea. These canopy heights were compared with the 1:5,000 scale forest maps used for the National Inventory Report in Korea. The GEDI data showed a maximum canopy height of 29.44 meters (m) in Gwangju, contrasting with the forest type maps that reported heights up to 34 m in Gwangju and parts of Icheon, and a minimum of 2 m in Icheon. Additionally, this study utilized Ordinary Least Squares(OLS)regression analysis to compare GEDI RH100 data with forest stand heights at the eup-myeon-dong level using ArcGIS, revealing Standard Deviations (SDs)ranging from -1.4 to 2.5, indicating significant regional variability. Areas where forest stand heights were higher than GEDI measurements showed greater variability, whereas locations with lower tree heights from forest type maps demonstrated lower SDs. The discrepancies between GEDI and actual measurements suggest the potential for improving height estimations through the application of high-resolution remote sensing techniques. To enhance future assessments of forest biomass and carbon storage at the Tier 3 level, high-resolution, reliable data are essential. These findings underscore the urgent need for integrating high-resolution, spatially explicit LiDAR data to enhance the accuracy of carbon sink calculations in Korea.

Keywords

1. Introduction

In 2015, the “Paris Agreement” was adopted at the 21st Conference of the Parties (COP21) to reduce greenhouse gases (GHGs) associated with the global climate crisis. Unlike the Kyoto Protocol (1997) (Ourbak and Tubiana, 2017), the Paris Agreement is a United Nations Framework Convention on Climate Change (UNFCCC) agreement that imposes GHG reduction commitments on all 195 Parties (Rogelj et al., 2016). The goal of the Paris Agreement is to hold the global average temperature increase to well below 2°C above pre-industrial levels and to pursue efforts to limit the temperature increase to 1.5°C above pre-industrial levels. To this end, each country must voluntarily define its climate action in Nationally Determined Contributions (NDCs), including GHG reduction targets, which are regularly submitted and updated, along with the status of their implementation, based on transparency principles (Hof et al., 2017).

OGCSBN_2024_v40n2_203_2_f0001.png 이미지

Fig. 1. Study flow.

The Intergovernmental Panel on Climate Change (IPCC) has proposed that by 2030, CO2 emissions should be reduced by at least 45% below 2010 levels to limit the rise in global average temperature and that net-zero carbon emissions should be achieved by absorbing 100 billion–1 trillion tCO2 by 2050 (Jung et al., 2023). In response, Korea has set its 2030 NDCs to reduce emissions by 37% compared to a Business as Usual (BAU) scenario (Jeong et al., 2022). Subsequently, as a follow-up to the 2050 Carbon Neutrality Declaration, the reduction target was increased to 40% compared to 2018 GHG emissions, and this was submitted to the UNFCCC.

For the submission of NDCs and the reporting on the implementation status, emission projections should be presented transparently through the preparation of GHG Inventories (Yona et al., 2020). To this end, each of the Parties will need to make efforts to develop activity data and scientific methodologies to build this inventory. GHG inventories are divided into five major categories (Energy; Industrial Processes; Agriculture; Land Use, Land Use Change and Forestry; and Waste) according to the IPCC Guidelines(GL) (Amon et al., 2021). To achieve the NDCs, it is important to reduce GHG emissions from the energy and industrial process sectors, but the carbon absorption aspect, which can offset carbon emissions, also plays a very important role considering the decline in industrial competitiveness and the impact on the national economy (Jin, 2023).

The LULUCF sector, which uniquely absorbs and stores GHGs, is critical for achieving reduction targets (Svoboda et al., 2022). This sector’s emissions and absorptions are estimated using activity data and GHG emission and absorption factors (Arets et al., 2021). Activity data are based on the six land use categories (forest land, cropland, grassland, wetlands, settlements, and other land) defined by the IPCC GL and are constructed by dividing areas retained in and converted from a category according to the degree of land-use change approach. However, it is difficult to identify and implement data on the area of land use and land-use change that is activity data, leading to problems of data consistency and reliability (Yu et al., 2015). GHG emissions are estimated by dividing them into Tier 1, Tier 2, and Tier 3 depending on the data used, and the IPCC recommends that the kind and type of basic data available in each country be considered and used for the estimation. As the tier increases, the accuracy and reliability of the estimated emissions improve, with major developed countries aiming for Tier 3 (Peter et al., 2016).

To apply Tier 3 to the LULUCF sector, it is essential to secure reliable geospatial data and to construct a land-use change matrix considering time series. In some countries, including China and Argentina, land-use matrices are constructed by combining lower-tiered approaches or Tiers 1 and 2 due to limitations in the available data and materials (Petrescu et al., 2012). Examples of countries that have applied Tier 3 include New Zealand and Australia. In these countries, satellite imagery such as Landsat and SPOT is used to detect land-use changes in spatial and temporal detail and to create land-use matrices (Flood et al., 2013). In Korea, however, Tier 1 is mostly applied to the LULUCF sector, with Tier 2 only applied to certain forest areas (Park et al., 2016). For advanced estimation of carbon sinks, it is necessary to detect time series land-use changes based on satellite imagery and to develop advanced biomass estimation methods. In this study, pilot data are constructed using the National Aeronautics and Space Administration (NASA) GEDI. GEDI is a new spaceborne Light Detection and Ranging (LiDAR) instrument operating onboard the International Space Station (ISS) and has been collecting data since April 2019 (Dubayah et al., 2020).

This study aims to enhance the scientific foundation for improving carbon sinks based on LULUCF. Firstly, it analyzes the current status and management of LULUCF inventories. Secondly, it examines the utilization of spatial information and satellite imagery in national inventories. Lastly, it proposes the application of ‘GEDI LiDAR’ for calculating carbon sinks in forests as a means to enhance the effectiveness of existing satellite imagery. This involves comparing GEDI LiDAR data with tree height data from the ‘forest type map,’ which is currently used for carbon sink estimation in South Korea, and subsequently deriving a regression equation to draw implications through correlation analysis.

2. GHG Management and GL

2.1. Status of GHG Management

At COP3 in Kyoto, Japan, the Kyoto Protocol was adopted, setting out the GHG reduction commitments of developed countries (United Nations, 1998). The Kyoto Protocol defined six types of GHG and required Annex I countries to reduce their emissions by an average of 5.2% from 1990 levels. At that time, Korea was classified as a Non-Annex I country and did not have a reduction commitment. Later, at the COP21 held in Paris, France, the Paris Agreement was adopted, which engaged all nations, including developing countries, in a common effort starting in 2020 (Ghezloun et al., 2017). The Paris Agreement imposed specific binding obligations on all Parties through the submission of NDCs every five years and regular annual reporting (Table 1).

Table 1. Comparing the Kyoto Protocol with the Paris Agreement (Adapted by Mor et al., 2023)

OGCSBN_2024_v40n2_203_4_t0001.png 이미지

MRV: Measurement, Reporting, and Verification.

The estimation of GHG emissions can be based on IPCC GL or country-specific methodologies (IPCC, 2006). As a GHG Inventory systematically lists sources and sinks of GHG, serving as the basis for policy development, the creation and construction of a quantitative and scientific inventory are required.

2.2. IPCC GL for GHG Inventories

The IPCC has established GL for compiling an inventory to estimate GHG emissions and removals by country (IPCC, 2006). As a result, many countries, including the United States, Japan, Australia, and European countries, use the IPCC GL to estimate their own GHG emissions. The IPCC developed GL for GHG Inventories (1996 and 2006) and Good Practice Guidance (GPG 2000 and 2003) to provide a universal methodology for estimating GHG Inventories to be reported to the UNFCCC (Table 2).

Table 2. Changes in IPCC GHG Inventories GL (Adapted by Yu et al., 2019)

OGCSBN_2024_v40n2_203_4_t0002.png 이미지

AFOLU: Agriculture, Forests, and Other Land-Use.

The 1996 IPCC GL specified the scope for compiling National GHG Inventories (IPCC, 1996). At COP9 in Italy in 2003, the IPCC GPG for LULUCF was adopted, and it was decided to follow these GL for the preparation of the annual National Inventory Report (NIR) starting in 2005. Subsequently, at COP10 in Argentina in 2004, a recommendation was made for Annex I Parties to prepare an inventory accounting carbon from activities under Articles 3.3 and 3.4 of the Kyoto Protocol. In 2006, the IPCC updated the GL for the preparation of NIR by incorporating GPG 2000 and GPG for LULUCF. The five major categories of National GHG Inventories as defined by the IPCC GL and GPG 2000 are Energy, Industrial Processes, Agriculture, LULUCF, and Waste. Of these, the LULUCF sector is the only one that absorbs and stores GHGs, and the six land-use categories defined by GPG for LULUCF are forest land, cropland, grassland, wetlands, settlements, and other land (IPCC, 2003).

GHG emissions in the LULUCF sector are estimated based on land use data, i.e., activity data, and GHG emission and absorption factors. The IPCC provides three approaches for Land-Use Change (LUC) for estimating the precise area and constructing an LUC matrix. The approaches are divided into Tiers 1 to 3 depending on the data used to estimate GHG emissions, and the Tiers are used in combination depending on the circumstances of each country (IPCC, 2006). The Tier 1 approach utilizes basic area identification and default emission factors for each land type provided by the IPCC GL. It requires the lowest amount of data, but uncertainty increases if the basic parameters available for each country and the suitability of the approach to its situation are unknown. The Tier 2 approach reflects the country’s current situation to apply LUC trends and country-specific emission factors.

The Tier 3 approach considers time series changes in the land use category and provides the most precise estimates but requires high-quality authoritative data. As GHG estimation methods become more sophisticated, the accuracy and reliability of emissions data improve; thus, countries around the world, including developed countries, are seeking to adopt the Tier 3 approach (Smith et al., 2012). Detecting LUCs over time is essential for estimation at the Tier 3 level. This study aims to analyze examples of building time-series and space-based land-use matrices used in developed countries and to suggest ways to advance carbon sink estimation methods.

3. Measures for Improved Estimation of Carbon Sinks in the LULUCF Sector

3.1. Status of Carbon Sink Estimation in the LULUCF Sector

To apply the Tier 3 approach, considering time-series changes in the LULUCF sector, it is essential to build a land-use change matrix. In examining the status of LUC matrix construction targeting the NIR of the Annex I Parties (38 countries), it was found that 60.5% (23 countries) of the Parties used remote sensing technology and data products (Mahdavi et al., 2018). Among these, the cases of LUC matrix construction in Japan, the United Kingdom, New Zealand, and Australia, where the Tier 3 approach is applied, were analyzed.

Japan’s GHG Inventory follows the 2006 IPCC GL. In the LULUCF sector, the Tier 3 approach is used to estimate CO2 emissions from forest land, cropland, and grassland (Ministry of the Environment Government of Japan, 2002). Building a LUC matrix requires various geospatial information in addition to basic statistical data. Japan uses remotely sensed data to estimate the area of each land type and estimates the area of forest cover change by comparing orthophotos with the latest imagery from satellites, including SPOT (Iizuka and Tateishi, 2015).

The United Kingdom derives land use and LUC values using the Report for Land Cover Map (LCM) 2007, a method that uses national statistical data and imagery information. The Report for LCM 2007 used imagery from satellites including Landsat-TM5, IRS-LISS 3, SPOT 5, and AWIFS. A land parcel specification with a Minimum Mappable Unit (MMU) of 0.5 ha and a Minimum Feature Width (MFW) of 20 m was displayed in the imagery information. To calculate area based on time series data, area data from 1970–1980 and 1980–1990 are used to estimate the area (Rowland et al., 2017).

New Zealand uses a combination of methods at all levels, including Tiers 1, 2, and 3, to calculate emissions and removals in the LULUCF sector. Tier 2 and 3 approaches are used to estimate carbon in grasslands with biomass, including pre-1990 natural forests and post-1989 natural forests, for which data are constructed using remotely sensed imagery (Fitzharris and McAlevey, 1999). Carbon absorption is measured using Landsat 4 and 5 and SPOT 5 data, and LUC information is collected using the forest carbon modeling and remotely sensed data in the Land Use and Carbon Analysis System (LUCAS). Land use maps are produced using wall-to-wall mapping, and the satellite imagery used is acquired only during the summer to extend the dynamic range of signals received from the ground (Scott et al., 2002).

In Australia, GHG emissions and removals are estimated using the Australian Greenhouse Gas Emissions Information System (AGEIS) and the Full Carbon Accounting Model (FullCAM) (Brack et al., 2006). FullCAM uses key spatial data, including land clearing, afforestation, and natural regeneration, derived from Landsat satellite imagery. National forest cover monitoring is conducted annually using Landsat satellite data (Multispectral Scanner, Thematic Mapper, Enhanced Thematic Mapper Plus Sensors, etc.). These data are created annually at a resolution of 25 meters, and geospatial information from the Australian Collaborative Land Use Mapping Program (ACLUMP) of the Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) is used as supplementary data.

Korea began securing systematic GHG inventory statistics with the enactment of the “Framework Act on Low Carbon, Green Growth” in 2010 (Yu et al., 2015). Based on this, the overall accuracy of GHG inventory estimation has been improved. However, due to a shortage of most data on land-use change in the LULUCF sector and the lack of integration of detailed management by element, the estimation is made at the Tier 1 level (Moon et al., 2021). For forest land, the change in carbon stock in living biomass is estimated at the Tier 2 level, but this estimation is unreliable as it is an indirect result of combining the emission equations with national geospatial data (e.g., forest type map). Despite the availability of internationally credible and up-to-date data and the technology to process such data in various ways, there is no policy basis for their use. In Western Europe and Japan, where the Tier 3 approach is used to estimate forest land in the LULUCF sector, satellite geospatial information including space-borne LiDAR GEDI; Landsat; etc., data is used to quantify carbon sinks and improve the accuracy of the estimates. For Korea to achieve its NDCs and carbon neutrality, it is necessary to conduct research on advanced carbon sink estimation and utilize internationally credible satellite geospatial data.

Table 3. Countries using the Tier 3 approach for estimating carbon sinks in the LULUCF sector and data used

OGCSBN_2024_v40n2_203_6_t0001.png 이미지

3.2. Carbon Sink Estimation Methods Based on Remote Sensing

Among the carbon sinks, forests play an important role in reducing atmospheric carbon concentrations in climate change by absorbing atmospheric CO2 and storing it as biomass (Besnard et al., 2021). In the estimation of forest biomass and carbon storage, accurate measurements of tree height and canopy height, along with periodic monitoring, are necessary (Alexander et al., 2018).

Methods for measuring forest height can be broadly categorized into field surveys and remote sensing. Field surveys are limited due to their high cost, time consumption, and the inaccessibility of some areas (St-Onge and Achaichia, 2001). Consequently, research on estimating forest height using remote sensing technology, which allows for the easier acquisition and processing of both 2-dimensional (Horizontal Structure) and 3-dimensional (Horizontal and Vertical Structure) vegetation structure data, is being conducted across wide areas. For small-scale forest areas (Local Scale), airborne LiDAR, radar, and high-resolution satellite imagery are employed to estimate tree height (Kugler et al., 2014), while for large-scale forest areas (Regional/Global Scale), low-resolution satellite imagery data such as Moderate Resolution Imaging Spectroradiometer (MODIS) and Landsat are utilized (Potapov et al., 2021).

Particularly, LiDAR, an active remote sensing system that acquires information by emitting radar waves onto the ground, is capable of capturing 3-dimensional vertical information on trees and vegetation structures, thereby reducing labor and time. It is extensively used for estimating forest cover (Lefsky et al., 2001). Recently, to enhance the accuracy of low-resolution satellite image data, research is actively being pursued to develop forest height estimation models through the fusion of spaceborne LiDAR GEDI data with field survey data (Qi et al., 2019).

The spaceborne LiDAR GEDI, one of the Earth Ventures Instruments operated by NASA, is a LiDAR system designed for forest observation. Mounted on the ISS, GEDI collects global data within the latitudinal range of 51.6°N to 51.6°S (Wang et al., 2013). GEDI data estimates the vertical structure and biomass of forest ecosystems, providing information with a resolution of a 25 m footprint (Dubayah et al., 2020). It offers detailed insights into forest structure, including tree height, branch density, and the vertical and horizontal distribution of leaves at a global level, which were missing from field surveys (Potapov et al., 2021).

In Korea, calculations of forest biomass and carbon storage are currently partially conducted at the Tier 2 level but fail to consider time series and regional characteristics, necessitating a Tier 3 level approach. This study proposes the measurement of tree height and canopy height using spaceborne LiDAR GEDI as a method to advance the calculation of biomass and carbon storage in forests, which serve as carbon sinks.

4. Applying the Improved Estimation of Carbon Sinks

4.1. GEDI Canopy Height Metrics

Knowing vertical forest structure is vital for assessing ecosystem carbon storage, annual productivity, and biodiversity (Fischer et al., 2019). Forest vertical structure plays a key role in modeling ecosystem functioning (Meyer et al., 2020). The RH with LiDAR to estimate biomass and canopy height. The RH is defined as the distance between the elevations of detected ground return and the accumulated waveform energy, where n ranges from 1 to 100 (García et al., 2011).

GEDI data provide information on the vertical distribution of vegetation by recording the Full Waveform LiDAR reflected by the height of vegetation. In this study, GEDI Level 2 (L2) data were collected and processed for Icheon, Gwangju, and Yeoju in Gyeonggi-do, Korea, to estimate the height of the forest canopy. GEDI L2 data are divided into L2A and L2B categories, providing information on the coordinates of each observation, forest and terrain heights, relative heights, as well as data on Canopy Cover Fraction and Leaf Area Index (Dubayah et al., 2020).

The data collected span from April 2019 to July 2021, and information was obtained for all pulses within the study area (Fig. 2). The data represent waveform return metrics for each 25 m diameter GEDI footprint. For each footprint, we extracted a set of the RH metrics RH50, RH90, RH95, and RH100 corresponding to the 50th, 90th, 95th, and 100th percentile of energy return height relative to the ground. These data are based on the L2A processing algorithm that uses different settings to control the waveform interpretation (Oliveira et al., 2023). For processing GEDI L2A data, preprocessing was performed using the data quality flags provided in the GEDI User Guide. We excluded minimum and maximum values and calculated the mean of the remaining four values to obtain the RH metric value for each footprint. We extracted geolocation information (latitude and longitude), forest height (RH 50, 95, 100), and ground elevation of GEDI pulses in the study area.

OGCSBN_2024_v40n2_203_7_f0001.png 이미지

Fig. 2. GEDI pulse locations in the study area.

4.2. Mapping Canopy and Forest Stand Height

A Canopy height map was created based on processed GEDI L2A data, which spatialized information on vegetation height (Fig. 3). Data for forest stand heights RH50, RH90, RH95, and RH100 were acquired from the location information of GEDI pulses, reflecting the forest status of the study area. To address missing values, the kriging interpolation method, commonly used in LiDAR studies (Fayad et al., 2022), estimated the values for areas with missing data at a spatial resolution of 100 meters. The Root Mean Square Standardized (RMSE) for RH50 was 0.90, while for RH90 and RH95, it was 0.91, and for RH100, it was 0.92, showing that the kriging of RH100 had the highest accuracy. The results of creating a canopy height map using GEDI data are as follows: heights ranged from 4.21 m to 29.45 m (Table 4). Gonjiam-eup in Gwangju City, Gyeonggi Province, recorded the highest canopy height of 29.44 m with an average height of 17.7 m. The lowest canopy heights were observed in Jeomdong-myeon, Yeoju City, ranging from a minimum of 4.21 m to a maximum of 25.37 m and an average of 12.35 m. The high canopy heights in the Gwangju area, where many GEDI pulses were measured, suggest that GEDI effectively reflects the current forest conditions. A comparative analysis will follow, using the canopy heights currently employed in Korea’s forest maps.

OGCSBN_2024_v40n2_203_8_f0001.png 이미지

Fig. 3. GEDI RH100 map in the study area.

Table 4. Average GEDI RH100 and forest stand height by eup-myeon-dong

OGCSBN_2024_v40n2_203_9_t0001.png 이미지

OGCSBN_2024_v40n2_203_10_t0001.png 이미지

Forestry data in Korea are provided by the Korea Forest Service in the form of 1:5,000 scale forest type maps displayed as polygons, which include various attribute information such as tree species and tree heights. However, there are high uncertainties in the estimates due to the provision of stand heights for each polygon and the limited use of field survey data (Lee et al., 2022). Fig. 4 shows data rasterized at a 100 m spatial resolution, excluding null values and heights of 0m within the forest type map. The observed stand heights range from a minimum of 2 m to a maximum of 34 m, with the highest heights recorded in Dochek-myeon, Gonjiam-eup in Gwangju City, and Sindun-myeon in Icheon City, showing maximum heights of 34m and average heights of 20.21 m, 15.92 m, and 19.97 m, respectively (Table 4). The lowest height was 2 m with an average of 14.17 m, observed in Hobeop-myeon and Yul-myeon in Icheon City.

OGCSBN_2024_v40n2_203_9_f0001.png 이미지

Fig. 4. Forest stand height map in the study area.

When comparing these measurements with the RH100 heights measured by GEDI, the high heights in Gonjiam-eup in Gwangju City were consistent. However, the highest values in Docheok-myeon were 26.79 m, which is 7.21 m lower, and in Sindun-myeon in Icheon City, there was a difference of 6.76 m. Additionally, the minimum height in Hobeop-myeon was 3.26 m higher, and in Yul-myeon, there was a difference of 5.87 m. These differences suggest limitations in data collection through field measurements and differences in resolution between aerial images, indicating a need for the adoption of space-borne LiDAR to construct high-resolution vegetation height maps for improved height measurement.

4.3. Regression Analysis of RH100 and Forest Stand Height

This study conducted an OLS regression analysis at the eup-myeon-dong level using the ArcGIS program to compare GEDI RH100 with forest stand height. OLS is a type of linear least squares method for estimating unknown parameters in a linear regression model (Ahmad et al., 2021). The dependent variable was forest stand height, and the independent variable was GEDI RH100 data. The analysis revealed that the range of Standard Deviation (SD) for OLS was between –1.4 and 2.5, showing regional variations in SD. Areas such as Docheok-myeon, Sinhyeon-dong, and Nongpyeong-dong in Gwangju City, and Baeksa-myeon and Bubal-eup in Icheon City, where forest stand height map heights were higher than GEDIRH100, showed higher variability with SDs ranging from 0.85 to 1.9.

OGCSBN_2024_v40n2_203_11_f0001.png 이미지

Fig. 5. Regional regression analysis of RH100 and forest stand height.

In contrast, areas like Buknae-myeon and Gangcheon-myeon near Yeoju City, Chowol-eup in Gwangju City and Yul-myeon in Icheon City, where the forest type map’s tree heights were lower, showed SDs from –1.4 to –0.81. Most other areas showed similar figures with SDs ranging from –0.80 to 0.29. This indicates that there is minimal discrepancy between GEDI measurements and the actual measured heights of the forest type heights. The significant differences in SDs in some areas can be attributed to differences between the spatially interpolated values based on the point location information of GEDI pulses and the actual measured heights. Therefore, if unmeasured values from field surveys could be further analyzed using high-resolution remote sensing techniques, more precise height measurements could be achieved.

5. Conclusions and Implications

In the 1990s, the phrase “climate change” was popularized, but by the 2020s this description had shifted to “climate crisis.” In other words, it took about 30 years for the international community to recognize climate change as a climate crisis—a very short time. In response, the international community adopted the Kyoto Protocol (1997) and the Paris Agreement (2015), proposing to achieve net-zero carbon emissions by 2050. Under the Paris Agreement, all Parties are required to prepare NDCs and submit NIR regularly and are obliged to present emissions projections transparently through the preparation of GHG Inventories. Therefore, for the international community to achieve net-zero carbon emissions, more accurate estimates of GHG Inventories are essential. The role of carbon sinks is very important in achieving NDCs, and in particular, data on carbon sinks, including forests and wetlands, can be obtained from the LULUCF sector.

Carbon emissions and absorption are estimated using activity data and GHG emission and absorption factors based on IPCC GL. However, it is difficult to identify and implement as data the area of land use and land-use change that is activity data, leading to problems of data consistency and reliability, and requiring high-level estimation approaches according to IPCC GL. The higher the level of estimation, also referred to as the Tier, the better it reflects the characteristics of a nation and region and considers spatial areas and time series. In Korea, for the estimation in the LULUCF sector, a Tier 2 approach is only applied to the forest land category, while other categories are estimated at the Tier 1 level due to lack of data and non-integrated management. However, countries using the Tier 3 approach estimate time series and area changes in land use using satellite geospatial information, including LiDAR, Landsat, and SPOT data. To improve carbon sink estimation, it is necessary to use internationally credible satellite geospatial information.

LiDAR data can provide information on the vertical structure of forests by measuring ground elevation. In particular, space-borne LiDAR can make observations from local to global scales and provide missing vegetation height information, leading to enhanced land cover observations from Landsat and MOIDS, and direct estimation of carbon absorption. In this study, space-borne GEDI LiDAR data were utilized to estimate forest canopy heights in the cities of Icheon, Gwangju, and Yeoju in Gyeonggi Province, Korea. L2A data were collected, and preprocessing was conducted to extract information on the geographical location of GEDI pulses, forest canopy height, and topographic elevation. Based on this data, RH100 was implemented at a 100 m spatial resolution using Kriging interpolation. Additionally, forest stand heights from current forest type maps used for carbon sink estimation in Korea was extracted and compared with the GEDI RH100 data. The analysis showed that the highest height measured by GEDI was 29.44 m in Gonjiam-eup, Gwangju City, and the lowest was 4.21 m in Jeomdong-myeon, Yeoju City.

In contrast, the forest type maps recorded the highest heights of 34 m in Dochek-myeon, Gonjiam-eup, and Sindun-myeon in Icheon City, with the lowest heights of 2 m in Hobeop-myeon and Yul-myeon in Icheon City. Regression analysis there were regions with little variation in SD. Most areas, except near Docheok-myeon, Sindun-myeon, Buknae-myeon, and Gangcheon-myeon, showed a minimal error with SD ranges between -0.8 and 0.29. These findings suggest discrepancies due to the uncertainties in field measurements and resolution differences in remote sensing images used in forest type map data creation. Future assessments of forest biomass and carbon storage at the Tier 3 level will require the use of high-resolution, reliable data. Accurate and precise greenhouse gas inventories necessitate combined studies using spatial data that reflect various national characteristics, alongside GEDI and high-resolution satellite imagery. This research could serve as foundational data for precise estimations of carbon sinks in the future.

Acknowledgments

This paper was written following the research work “Study on the Carbon Sink Verification System Using Biomass Observed by Satellite” (RE2024-04) funded by the Korea Environment Institute (KEI).

Conflict of Interest

No potential conflict of interest relevant to this article was reported.

References

  1. Ahmad, I., Dar, M. A., Fenta, A., Halefom, A., Nega, H., Andualem, T. G., and Teshome, A., 2021. Spatial configuration of groundwater potential zones using OLS regression method. Journal of African Earth Sciences, 177, 104147. https://doi.org/10.1016/j.jafrearsci.2021.104147
  2. Alexander, C., Korstjens, A. H., and Hill, R. A., 2018. Influence of micro-topography and crown characteristics on tree height estimations in tropical forests based on LiDAR canopy height models. International Journal of Applied Earth Observation and Geoinformation, 65, 105-113. https://doi.org/10.1016/j.jag.2017.10.009
  3. Amon, B., Cinar, G.,Anderl, M., Dragoni, F., Kleinberger-Pierer, M., and Hortenhuber, S., 2021. Inventory reporting of livestock emissions: The impact of the IPCC 1996 and 2006 Guidelines. Environmental Research Letters, 16(7), 075001. https://doi.org/10.1088/1748-9326/ac0848
  4. Arets, E. J. M. M., Van Der Kolk, J. W. H., Hengeveld, G. M., Lesschen, J. P., Kramer, H., Kuikman, P. J., and Schelhaas, N. J., 2021. Greenhouse gas reporting for the LULUCF sector in the Netherlands: Methodological background, update 2021 (WOt-technical Report No. 201). WOT Natuur & Milieu. https://doi.org/10.18174/539898
  5. Besnard, S., Santoro, M., Cartus, O., Fan, N., Linscheid, N., Nair, R. et al., 2021. Global sensitivities of forest carbon changes to environmental conditions. Global Change Biology, 27(24), 6467-6483. https://doi.org/10.1111/gcb.15877
  6. Brack, C., Richards, G., and Waterworth, R., 2006. Integrated and comprehensive estimation of greenhouse gas emissions from land systems. Sustainability Science, 1, 91-106. https://doi.org/10.1007/s11625-006-0005-6
  7. Dubayah, R., Blair, J. B., Goetz, S., Fatoyinbo, L., Hansen, M., Healey, S. et al., 2020. The global ecosystem dynamics investigation: High-resolution laser ranging of the Earth's forests and topography. Science of Remote Sensing, 1, 100002. https://doi.org/10.1016/j.srs.2020.100002
  8. Fayad, I., Baghdadi, N., Bailly, J. S., Frappart, F., and Pantaleoni Reluy, N., 2022. Correcting GEDI water level estimates for inland waterbodies using machine learning. Remote Sensing, 14(10), 2361. https://doi.org/10.3390/rs14102361
  9. Fischer, R., Knapp, N., Bohn, F., Shugart, H. H., and Huth, A., 2019. The relevance of forest structure for biomass and productivity in temperate forests: New perspectives for remote sensing. Surveys in Geophysics, 40, 709-734. https://doi.org/10.1007/s10712-019-09519-x
  10. Fitzharris, B., and McAlevey, B., 1999.Remote sensing of seasonal snow cover in the mountains of New Zealand using satellite imagery. Geocarto International, 14(3), 35-44. https://doi.org/10.1080/10106049908542115
  11. Flood, N., Danaher, T., Gill, T., and Gillingham, S., 2013. An operational scheme for deriving standardised surface reflectance from Landsat TM/ETM+ and SPOT HRG imagery for Eastern Australia. Remote Sensing, 5(1), 83-109. https://doi.org/10.3390/rs5010083
  12. Garcia, M., Riano, D., Chuvieco, E., Salas, J., and Danson, F. M., 2011. Multispectral and LiDAR data fusion for fuel type mapping using Support Vector Machine and decision rules. Remote Sensing of Environment, 115(6), 1369-1379. https://doi.org/10.1016/j.rse.2011.01.017
  13. Ghezloun, A., Saidane, A., and Merabet, H., 2017. The COP 22 new commitments in support of the Paris Agreement. Energy Procedia, 119, 10-16. https://doi.org/10.1016/j.egypro.2017.07.040
  14. Hof, A. F., den Elzen, M. G., Admiraal, A., Roelfsema, M., Gernaat, D. E., and van Vuuren, D. P., 2017. Global and regional abatement costs of Nationally Determined Contributions (NDCs) and of enhanced action to levels well below 2 ℃ and 1.5 ℃. Environmental Science & Policy, 71, 30-40. https://doi.org/10.1016/j.envsci.2017.02.008
  15. Iizuka, K., and Tateishi, R., 2015. Estimation of CO2 sequestration by the forests in Japan by discriminating precise tree age category using remote sensing techniques. Remote Sensing, 7(11), 15082-15113. https://doi.org/10.3390/rs71115082
  16. IPCC, 1996. IPCC guidelines for national greenhouse gas inventories. Available online: https://www.ipcc-nggip.iges.or.jp/public/gl/invs1.html (accessed on Apr. 23, 2024).
  17. IPCC, 2003. Good practice guidance for Land Use, Land-Use Change and Forestry. Available online: https://www.ipcc.ch/site/assets/uploads/2018/03/GPG_LULUCF_FULLEN.pdf (accessed on Apr. 23, 2024).
  18. IPCC, 2006. 2006 IPCC Guidelines for National Greenhouse Gas Inventories. Available online: https://www.ipcc-nggip.iges.or.jp/meeting/pdfiles/Washington_Report.pdf (accessed on Apr. 23, 2024).
  19. Jeong, W.-C., Lee, D.-H., Roh, J. H., and Park, J.-B., 2022. Scenario analysis of the GHG emissions in the electricity sector through 2030 in South Korea considering updated NDC. Energies, 15(9), 3310. https://doi.org/10.3390/en15093310
  20. Jin, I., 2023. Probability of achieving NDC and implications for climate policy: CO-STIRPAT approach. Journal of Economic Analysis, 2(4), 82-97. https://doi.org/10.58567/jea02040005
  21. Jung, J. A., Kwon, B. O., Hong, H. J., Ahn, J. H, and Lee, M. J., 2023, Comparative analysis of blue carbon stock spatial data in the estuaries and coastal areas of the Geum and Nakdong Rivers. Korean Journal of Remote Sensing, 39(6), 1505-1515. https://doi.org/10.7780/kjrs.2023.39.6.1.28
  22. Kugler, F., Schulze, D., Hajnsek, I., Pretzsch, H., and Papathanassiou, K. P., 2014.TanDEM-XPol-InSAR performance for forest height estimation. IEEE Transactions on Geoscience and Remote Sensing, 52(10), 6404-6422. https://doi.org/10.1109/tgrs.2013.2296533
  23. Lee, B. N.,Jung, G. H., Rhy,J. Y.,Kwon, G.W., Yim,J. S. and Park, J. W., 2022, Review of remote sensing technology for forest canopy height estimation and suggestions for the advancement of Korea's nationwide canopy height map. Journal of Korean Society of Forest Science, 111(3), 435-449. https://doi.org/10.14578/jkfs.2022.111.3.435
  24. Lee, S. J., Yim, J. S., Son, Y. M., Son, Y., and Kim, R., 2018. Estimation of forest carbon stocks for national greenhouse gas inventory reporting in South Korea. Forests, 9(10), 625. https://doi.org/10.3390/f9100625
  25. Lefsky, M., Cohen, W., and Spies, T., 2001. An evaluation of alternate remote sensing products for forest inventory, monitoring, and mapping of Douglas-fir forests in western Oregon. Canadian Journal of Forest Research, 31(1), 78-87. https://doi.org/10.1139/x00-142
  26. Mahdavi, S., Salehi, B., Granger, J., Amani, M., Brisco, B., and Huang, W., 2018.Remote sensing for wetland classification: A comprehensive review. GIScience & Remote Sensing, 55(5), 623-658. https://doi.org/10.1080/15481603.2017.1419602
  27. Meyer, L., Diniz-Filho, J. A. F., Lohmann, L. G., Hortal, J., Barreto, E., Rangel, T., and Kissling, W. D., 2020. Canopy height explains species richness in the largest clade of Neotropical lianas. Global Ecology and Biogeography, 29(1), 26-37. https://doi.org/10.1111/geb.13004
  28. Ministry of the Environment Government of Japan, 2002.Japan's National Greenhouse Gas Emissions in Fiscal Year 2020 (Final Figures). Available online: https://www.env.go.jp/en/headline/2599.html (accessed on Apr. 4, 2024).
  29. Moon, G.-H., Yim, J.-S., and Moon, N.-H., 2021. Optimal sampling intensity in South Korea for a land-use change matrix using point sampling. Land, 10(7), 677. https://doi.org/10.3390/land10070677
  30. Mor, S., Aneja, R., Madan, S., and Ghimire, M., 2023. Kyoto Protocol and Paris Agreement: Transition from bindings to pledges - A review. Millennial Asia, 097639962211 41546. https://doi.org/10.1177/09763996221141546
  31. Oliveira, P. V., Zhang, X., Peterson, B., and Ometto, J. P., 2023. Using simulated GEDI waveforms to evaluate the effects of beam sensitivity and terrain slope on GEDI L2A relative height metrics over the Brazilian Amazon Forest. Science of Remote Sensing, 7, 100083. https://doi.org/10.1016/j.srs.2023.100083
  32. Ourbak, T., and Tubiana, L., 2017. Changing the game: the Paris Agreement and the role of scientific communities. Climate Policy, 17(7), 819-824. https://doi.org/10.1080/14693062.2017.1348331
  33. Park, S.-J., Lee, C.-H., Kim, M.-S., Yun, S.-G., Kim, Y. H., & Ko, B. G., 2016. Calculation of GHGs emission from LULUCF-cropland sector in South Korea. Korean Journal of Soil Science and Fertilizer, 49(6), 826-831. https://doi.org/10.7745/kjssf.2016.49.6.826
  34. Peter, C., Fiore, A., Hagemann, U., Nendel, C., and Xiloyannis, C., 2016. Improving the accounting of field emissions in the carbon footprint of agricultural products: A comparison of default IPCC methods with readily available medium-effort modeling approaches. The International Journal of Life Cycle Assessment, 21, 791-805. https://doi.org/10.1007/s11367-016-1056-2
  35. Petrescu,A.,Abad-Vinas, R., Janssens-Maenhout, G., Blujdea, V., and Grassi, G., 2012. Global estimates of carbon stock changes in living forest biomass: EDGARv4.3 - Time series from 1990 to 2010. Biogeosciences, 9(8), 3437-3447. https://doi.org/10.5194/bg-9-3437-2012
  36. Potapov, P., Li, X., Hernandez-Serna, A., Tyukavina, A., Hansen, M.C., Kommareddy, A., Pickens, A. et al., 2021. Mapping global forest canopy height through integration of GEDI and Landsat data. Remote Sensing of Environment, 253, 112165. https://doi.org/https://doi.org/10.1016/j.rse.2020.112165
  37. Qi, W., Saarela, S., Armston, J., Stahl, G., and Dubayah, R., 2019. Forest biomass estimation over three distinct forest types using TanDEM-XInSAR data and simulated GEDI lidar data. Remote Sensing of Environment, 232, 111283. https://doi.org/10.1016/j.rse.2019.111283
  38. Rogelj, J., Den Elzen, M., Hohne, N., Fransen, T., Fekete, H., Winkler H. et al., 2016. Paris Agreement climate proposals need a boost to keep warming well below 2 ℃. Nature, 534(7609), 631-639. https://doi.org/10.1038/nature18307
  39. Rowland, C., Morton, R., Carrasco, L., McShane, G., O'neil, A., and Wood, C., 2017. Land cover map 2015 (vector, GB). NERC Environmental Information Data Centre, 10. https://doi.org/10.5285/6c6c9203-7333-4d96-88ab-78925e7a4e73
  40. Scott, N., Tate, K., Giltrap, D., Smith, C. T., Wilde, H., Newsome, P. J. F., and Davis, M. R., 2002. Monitoring land-use change effects on soil carbon in New Zealand: Quantifying baseline soil carbon stocks. Environmental Pollution, 116, S167-S186. https://doi.org/10.1016/s0269-7491(01)00249-4
  41. Smith, P., Davies, C. A., Ogle, S., Zanchi, G., Bellarby, J., Bird, N. et al., 2012. Towards an integrated global framework to assess the impacts of land use and management change on soil carbon: current capability and future vision. Global Change Biology, 18(7), 2089-2101. https://doi.org/10.1111/j.1365-2486.2012.02689.x
  42. St-Onge, B. A., and Achaichia, N., 2001. Measuring forest canopy height using a combination of lidar and aerial photography data. International Archives of Photogrammetry Remote Sensing and Spatial Information Sciences, 34(3/W4), 131-138.
  43. Svoboda, J., Stych, P., Lastovicka, J., Paluba, D., and Kobliuk, N., 2022. Random forest classification of Land Use, Land-Use Change and Forestry (LULUCF) using Sentinel-2 data-A case study of Czechia. Remote Sensing, 14(5), 1189. https://doi.org/10.3390/rs14051189
  44. United Nations, 1998. Kyoto Protocol to the United Nations Framework Convention on Climate Change. Available online: https://unfccc.int/resource/docs/convkp/kpeng.pdf (accessed on Apr. 23, 2024).
  45. Wang, X., Ouyang, S., Sun, O. J., and Fang, J., 2013. Forest biomass patterns across northeast China are strongly shaped by forest height. Forest Ecology and Management, 293, 149-160. https://doi.org/10.1016/j.foreco.2013.01.001
  46. Yona, L., Cashore, B., Jackson, R. B., Ometto, J., and Bradford, M. A., 2020. Refining national greenhouse gas inventories. Ambio, 49, 1581-1586. https://doi.org/10.1007/s13280-019-01312-9
  47. Yu, S. C., Ahn, W., and Ok, J. A., 2015. A study on construction plan of the statistics for national greenhouse gas inventories (LULUCF Sector). Spatial Information Research, 23(3), 67-77. https://doi.org/10.12672/ksis.2015.23.3.067
  48. Yu, S. C., Shin, D. B., and Ahn, J. W., 2019. Analysis of Land Use, Land-use Change, and Forestry (LULUCF) and construction of statistics in Korea. Sensors & Materials, 31. https://doi.org/10.18494/sam.2019.2340