Background & Summary

Grasslands, covering approximately one-quarter of global land use, are pivotal in the carbon cycle, storing around 10% (343 Gt) of global soil organic carbon (SOC)1,2. This substantial carbon storage underscores their role as essential carbon sinks within terrestrial ecosystems3,4. Despite their importance, research on carbon storage and sequestration in South Korean grasslands remain limited, with previous studies primarily focused on agricultural and forested lands5.

Currently, SOC stocks in South Korean grasslands are estimated using the Tier 1 approach outlined in the Intergovernmental Panel on Climate Change Guidelines (IPCC GL) for National Greenhouse Gas Inventory6,7. This approach uses default reference SOC stock values provided in the IPCC GL for each soil type within a climate zone comparable to South Korea (warm temperate moist). However, relying on these generalized values, rather than country-specific measurements, fails to capture the unique characteristics of South Korean soils. This limitation highlights the urgent need for field-measured SOC data to develop country-specific reference SOC stock values for each soil types, accurately reflecting the diverse soils found in South Korean grasslands.

Another limitation in estimating SOC stocks in South Korean grasslands is the limited inclusion of management practices8. According to the current Greenhouse Gas Inventory method used in the grassland sector in South Korea, CO2 emission and absorption are only recorded when grassland area changes due to land-use conversion. However, even in the grasslands remaining grasslands, the intensity of management practices can influence SOC accumulation and/or stabilization, thereby affecting CO2 emission and absorption. For instance, management practices, such as fertilization, liming, and overseeding, not only boost forage yield but also enhance SOC accumulation9,10. These practices can contribute to carbon stabilization in grassland soils by increasing organic matter inputs or improving soil structural stability2,9,10. Thus, it is crucial to consider management intensity for more accurate and representative estimates of SOC stock in South Korean grasslands.

In this study, we provide a field-measured dataset of SOC stocks from 46 grassland sites across South Korea, accounting for variation in soil type and management intensity. The dataset comprises analysis results from 208 soil core samples, encompassing four soil types (high-activity clay, low-activity clay, sandy, and volcanic soils) and three levels of management intensity (improved, nominally managed, and degraded). Alongside SOC stock measurements, soil properties indicating of organic matter quality, such as C/N and H/C ratios, were analyzed. This dataset provides a foundational resource for accurately estimating SOC stocks in South Korean grasslands by developing country-specific reference SOC stock values and incorporating SOC changes based on management intensity. Furthermore, it will advance South Korea’s National Greenhouse Gas Inventory for grasslands to Tier 2 and Tier 3, aligned with the 2006 IPCC GL for the Agriculture, Forestry, and Other Land Use (AFOLU) sector.

Methods

Study site

This study was conducted at 46 grassland sites in South Korea, each maintained as grasslands for at least 20 years (Table S1). The sites are distributed across the following regions: 2 in Chungcheongbuk-do (CB), 8 in Chungcheongnam-do (CN), 1 in Gyeonggi-do (GG), 4 in Gyeongsangnam-do (GN), 16 in Gangwon-do (GW), 3 in Jeollabuk-do (JB), 10 in Jeju (JJ), and 2 in Jeollanam-do (JN) (Fig. 1a). The 30-year mean temperature and annual precipitation (1991–2020) were derived from 1/24° (~4 km) spatial resolution monthly global dataset, TerraClimate11 (Table S1).

Fig. 1
figure 1

Sampling sites and sample description: (a) Map of sampling sites (total sites = 46), (b) soil sample distribution by soil type, and (c) soil sample distribution by management intensity.

Full size image

Soil types were classified into four groups-volcanic, sandy, high-activity clay (HAC), and low-activity clay (LAC)-based on the United States Department of Agriculture (USDA) Soil Taxonomy12, an approach also recognized by the 2006 IPCC GL6. The soil type classification process followed a hierarchical decision-making framework, as illustrated in Fig. 2. The initial step involved identifying “volcanic” soils characterized by low bulk density (BD ≤ 0.9 g cm−3), moderate organic carbon content (OC ≤ 25%), high phosphate (P) retention (≥85%), and high acid oxalate (Alo + ½ Feo) content (≥2%). Soils that did not meet the volcanic criteria were subsequently evaluated based on their textural composition. Specifically, soils with sand content exceeding 70% and clay content below 8% were classified as “sandy” soils. For the remaining soils, the cation exchange capacity (CEC) relative to clay content was assessed to distinguish between high-activity and low-activity clay soils. Soils with a CEC-to-clay ratio of ≥0.3 were categorized as “HAC” soils, while those with a lower ratio were designated as “LAC” soils. The spatial distribution of soil types across study sites is presented in Fig. 1b.

Fig. 2
figure 2

Classification of Soil Types Using a Decision Tree Framework: Soil types were classified into four groups, including volcanic, sandy, high-activity clay (HAC), and low-activity clay (LAC) soils. The classification was based on bulk density (BD), organic carbon (OC), phosphate (P) retention, acid oxalate (Alo + 1/2Feo) content, soil texture, and cation exchange capacity (CEC).

Full size image

We adjusted the four-level management intensity categories (FMG) from the 2006 IPCC GL6 into three levels (improved, nominally managed, degraded) to align with the existing grasslands survey conducted in South Korea13,14 (Table 1). The only nationwide grasslands management survey available in South Korea is the ‘National Grasslands Management Survey’, conducted by the Ministry of Agriculture and Food. This survey classifies grasslands into three grades based on % grass cover. Consequently, we assigned ‘upper grade’ grasslands to ‘improved’, ‘middle grade’ to ‘nominally managed’, and ‘low grade’ to ‘degraded’. Improved grasslands correspond to the ‘upper grade’ category of the National Grassland Management Survey, indicating grass cover of 80% or more. Grasslands in this category typically receive active management, including controlled grazing intensity and at least one or more management interventions per year, such as fertilization, irrigation, and grass overseeding. Nominally managed grasslands correspond to the ‘middle grade’, characterized by grass cover ranging from 60% to 80%. These grasslands typically have moderate grazing intensity and receive only one management intervention per year, specifically fertilization. Degraded grasslands correspond to the ‘low grade’, representing grasslands with grass cover below 60%. Grasslands in this category typically have uncontrolled grazing intensity and do not receive fertilization. The distribution of our study sites by management intensity is presented in Fig. 1c.

Table 1 Classification criteria for management intensity, as outlined in the 2006 IPCC Guidelines (IPCC GL), grasslands grading criteria from the National Grasslands Management Survey in South Korea, and integrated management intensity categories applied in this study.
Full size table

Soil samples

Soil sampling was performed between April and October from 2020 to 2024. A total of 207 soil core samples were collected from 45 sites, with 3 to 9 samples per site, depending on the field area. However, only one soil core sample was collected from site JJ06 due to its limited area. Following the 2006 IPCC GL6, soil samples were collected at 0-30 cm depth using a soil core sampler (10 cm diameter; 2” × 12” Soil Core Sampler Complete, AMS, American Falls, ID, USA). Each core was divided into surface (0-15 cm) and subsurface (15-30 cm) layers to assess SOC distribution by depth. Soil samples were air-dried for two 2 weeks before proceeding with further soil analyses.

Soil analysis

Soil analyses were conducted to determine soil texture, gravel ratio, bulk density, and total carbon, nitrogen, and hydrogen content. Soil texture was evaluated following methods described in the Kellogg Soil Survey Laboratory Methods Manual12. The gravel ratio was calculated by sieving air-dried soil through a 2 mm mesh, with the fraction retained on the sieve considered as gravel and expressed as a percentage of the total soil weight. Soil bulk density was determined by calculating the ratio of the soil oven-dry weight at 105°C to the soil core volume15. For total C, N, and H analyses, air-dried soil samples passing a 2-mm sieve were finely ground using a ball mill (MM400 Ball Mill; Retsch GmbH, Haan, NRW, Germany) and stored in sterile 2-mL tubes. These prepared samples were analyzed via combustion (900 °C) using an elemental analyzer (Flash EA 1112; Thermo Fisher Scientific, Waltham, MA, USA). Consequently, SOC stocks for each layer were calculated using the gravel ratio, bulk density, and total carbon content specific to that layer, as detailed in the following section, Soil organic carbon stock (Fig. 3). Additionally, C/N and H/C ratios were calculated to assess soil organic matter quality (Figs. 4, 5).

Fig. 3
figure 3

Soil organic carbon (SOC) stock: (a,d) SOC stock at 0-15 cm and 15-30 cm depths across all 46 sites, (b,e) SOC stock at 0-15 cm and 15-30 cm depths by soil type, and (c,f) SOC stock at 0-15 cm and 15-30 depths by management intensity. Statistical significance is indicated as **(p < 0.05).

Full size image
Fig. 4
figure 4

Soil C/N ratios: (a,d) Soil C/N ratio at 0-15 cm and 15-30 cm depths across all 46 sites, (b,e) soil C/N ratio at 0-15 cm and 15-30 cm depths by soil type, and (c,f) soil C/N ratio at 0-15 cm and 15-30 depths by management intensity. Statistical significance is indicated as **(p < 0.05).

Full size image
Fig. 5
figure 5

Soil H/C ratios: (a,d) Soil H/C ratio at 0-15 cm and 15-30 cm depths across all 46 sites, (b,e) soil H/C ratio at 0–15 cm and 15-30 cm depths by soil type, and (c,f) soil H/C ratio at 0-15 cm and 15-30 depths by management intensity. Statistical significance is indicated as **(p < 0.05).

Full size image

Soil organic carbon stock

The SOC stocks were calculated using the following equation recommended in the 2006 IPCC GL6:

$${{SOC\; stock}}_{i}=\frac{{B}_{i}\times {C}_{i}\times \left(\frac{100-{G}_{i}}{100}\right)\times {D}_{i}}{10}$$

where, Bi represents the bulk density at depth i (g cm−3), Ci is the total carbon content at depth i (g C kg−1), Gi denotes the gravel ratio at depth i (%), and Di is the thickness of soil layer at depth i (cm). This equation calculates SOC stocks in tons of carbon per hectare (tC ha−1) by incorporating bulk density, total carbon content, and gravel ratio for each specified depth layer.

Statistical analyses

Statistical analyses were conducted using SAS software (SAS Institute Inc., Cary, NC, USA). Analysis of variance (ANOVA) was conducted using the ‘PROC GLM’ procedure to examine the effects of soil type on SOC stock, C/N, and H/C ratios, and multiple comparisons were performed using least square means (LSMeans) at significance levels of p < 0.05 and p < 0.10. The SAS script and detailed workflow are provided in the Supplementary File (Tables S2, S3).

Data Records

The dataset is available at Figshare (https://doi.org/10.6084/m9.figshare.27935013.v2)16 and (https://doi.org/10.6084/m9.figshare.27935025.v1)17. This section provides a detailed description of the dataset’s contents and organization. The dataset consists of two files:

South Korean Grasslands_Study Sites

This file describes the information of the sites where the soil cores have been collected, including site ID, year of establishment, coordinates (latitude and longitude), sampling time, mean temperature and annual precipitation (1991–2020), pH, EC, texture, soil types, and management intensity16.

South Korean grasslands_soil analysis results

This file provides gravel ratio, bulk density, total carbon, nitrogen, and hydrogen contents, C/N and H/C ratios, and SOC stock17, which are measured and calculated using the method as described in the sections, Methoud-Soil analysis and Soil organic carbon stock.

Each dataset is structured to support reproducibility and further comparative analyses of SOC stock dynamics in South Korean grasslands.

Technical Validation

To ensure the validity of this dataset, we evaluated (1) the representativeness of the dataset for South Korean grasslands, and (2) the reliability of the measured SOC stock, C/N, and H/C data.

Representativeness of the dataset for South Korean grasslands

To assess the dataset’s representativeness, we compared our site distribution with national surveys on grassland areas and soil types. According to national survey data13, grasslands in South Korea cover approximately 31,784 ha, predominantly located in Jeju-do (48.6%) and Gangwon-do (15.6%). Our dataset aligns closely with this national distribution, with 56.5% of our sampling sites located in Jeju-do and Gangwon-do. Additional sites were strategically selected across Chungcheong-do, Gyeonggi-do, Gyeongsang-do, and Jeolla-do to represent diverse environmental and climatic conditions across the country. Furthermore, according to the 2023 National Greenhouse Gas Inventory Report8, the predominant soil type in South Korean grasslands is HAC, accounting for approximately 72.0% of the total grassland soil area. This is followed by volcanic soils (18.5%), sandy soils (5.5%), and LAC soils (4.1%). The distribution of soil types in our dataset follows the same order as that in national statistics, supporting the dataset’s representativeness.

While our site distribution concerning grasslands management intensity (improved: 28.3%, nominally managed: 65.2%, degraded: 6.5%) differs from the National Grassland Management Survey results (upper grade: 30.2%, middle grade: 33.4%, low grade: 36.4%)17, this divergence arises from our sampling strategy, which intentionally prioritized nominally managed grasslands. This strategic decision aligns with the 2006 IPCC GL, which recommends nominally managed grasslands as the baseline category for deriving SOC reference values across soil types6. Consequently, although our dataset effectively represents nominally managed conditions, additional sampling of improved and degraded grasslands would further enhance representativeness across all management intensities.

Our dataset, in terms of physicochemical characteristics, represented acidic soils with an average pH of 5.7 ± 0.9 and relatively low electrical conductivity (EC) levels of 0.07 ± 0.06 dS m−1 (Table S1). These values are indicative of typical temperate grassland soils characterized by non-saline conditions18,19.

Reliability of SOC stock, C/N, and H/C data

Figures 3–5 illustrate the distribution of SOC stock, C/N, and H/C ratios across the 46 grassland sites. The results of ANOVA analysis for SOC stock, C/N, and H/C ratios are provided in Supplementary File (Table S3). SOC stock values in 0-15 cm and 15-30 cm depth ranged from 1.5 to 198.0 tC ha-1 (mean: 33.5 tC ha-1) (Fig. 3), aligning with previously reported ranges for grassland soils, such as Phukubye et al. (2022) (0.8-127.0 tC ha-1, mean: 30.2 tC ha-1)20 and Eze et al. (2018) (1.8–200.8 tC ha-1, mean: 43.4 tC ha-1)9. Management intensity played a key role in SOC variation, with improved grasslands exhibiting the highest SOC stocks (mean: 41.0 tC ha-1), while degraded sites had the lowest (mean: 21.4 tC ha-1; Fig. 3c,f). These findings align with Phukubye et al. (2022), who analyzed 235 global grassland sites and reported that improved management increased SOC stocks by as much as 18 tC ha-1 compared to unmanaged conditions20. C/N ratios in this study ranged from 2.9 to 46.2 (mean: 10.2; Fig. 4), which are consistent with the global range reported for grassland soils20, supporting the robustness of the dataset. In contrast, H/C ratio values for grassland soils have not been previously documented, limiting direct comparisons. However, the values obtained in this study provide a valuable baseline for assessing organic matter composition and decomposition dynamics (Fig. 5). Future research should explore H/C ratio variability under different management practices to enhance understanding of soil carbon cycling.

To ensure analytical accuracy and reproducibility, the elemental analyzer was regularly calibrated using certified reference materials (Low Organic Content Soil Standard OAS, Elemental Microanalysis, Okehampton, UK), and duplicate measurements were conducted to validate precision. The strong agreement between the dataset and previously reported values reinforces this study’s methodological reliability and classification framework.

Usage Notes

This dataset provides a foundational resource for accurately estimating SOC stocks in South Korean grasslands, offering a basis for developing a country-specific SOC coefficient aligned with IPCC LULUCF standards. Soil samples were collected and analyzed using verified methods, and SOC stocks were precisely estimated while accounting for variations in soil types and management intensities. Accordingly, this field-measured dataset can improve greenhouse gas inventory estimates in South Korean grasslands to Tier 2 under the 2006 IPCC GL. Furthermore, it can serve as model input data, potentially enabling a Tier 3 approach, thereby supporting carbon-neutral policies and carbon offset projects focused on soil carbon preservation and sustainable grasslands management.

However, several limitations should be considered when applying this dataset. First, possible sources of measurement error include sampling’s spatial and temporal representativeness. Regarding spatial representativeness, although sampling points were strategically selected to evenly represent the dominant regional grass species and 3 to 9 replicate soil core samples were collected per site depending on the field area, variations in grazing intensity, fertilization, irrigation, and overseeding practices across grassland farms could introduce additional uncertainties that are not fully captured by our approach. Temporal representativeness presents another potential source of measurement error. All soil sampling was conducted at the peak of aboveground grass biomass production to mitigate intra-annual fluctuations. Despite SOC stock typically being a slow-changing variable, occasional seasonal variations have been reported21, meaning pronounced seasonal variability in temperate grasslands could introduce further temporal uncertainties. Additionally, uncertainties in SOC stock calculations arise from inherent site-specific variability in gravel content, influenced by factors such as parent material and soil formation processes. Gravel content varied significantly among our sampling sites, ranging from 0.1% to 76.0% (mean: 24.1%), affecting the accuracy of SOC stock estimations. Users are therefore advised to carefully account for this inherent variability when interpreting and utilizing the dataset.