Towards a remote sensing-based assessment of carbon emissions from peatlands

Towards a remote sensing-based assessment of carbon emissions from peatlands Pouya Ghezelayagh1, Andrzej Kamocki2, Piotr Banaszuk2 & Mateusz Grygoruk1 Among greenhouse gases-generating sources, biosphere sources from natural carbon (C) reservoirs play a significant role. A vital component of the biosphere is peatlands—the largest natural terrestrial carbon storage on the earth. Peatlands function as both C sink and C source, showing their pivot role in mitigating GHGs. Releasing C results from peat oxidation—the decomposition of organic matter in the peat. This decomposition reduces the volume of peat and, hence, causes subsidence. This study introduces an exclusive remote-sensing-based framework for estimating carbon emissions from peatlands using subsidence rates. This framework integrates peat properties—bulk density and soil organic carbon—with the oxidated peat subsidence, which refers to the proportion of subsidence attributed to the oxidation process rather than shrinkage. Achieving a fully remote-sensing-based approach promises time-effective, cost-effective, and consistent C emission monitoring even in unreachable places in peatlands, addressing the critical need for global climate change mitigation strategies. However, this achievement requires collaborative efforts among researchers to implement it in other sites to improve dataset accuracy for each parameter. By improving this framework, the scientific community can pave the way for robust, large-scale assessments of peatland C emission. Keywords Peat degradation, Subsidence, Soil organic carbon (SOC), Bulk density, Oxidation, Climate change Climate change has risen as one of the highest global concerns in contemporary times1,2. The increasing levels of greenhouse gases (GHGs) in the atmosphere have resulted in ascending temperatures and many damaging environmental consequences3–5. GHGs are atmospheric components that inhibit the escape of the Sun’s reflected electromagnetic radiation from Earth’s surface back into space6,7. These gases—primarily carbon dioxide (CO₂), methane (CH₄), nitrous oxide (N₂O), and water vapor (H₂O)—function analogously to the glass walls of a greenhouse, trapping heat within the atmosphere8,9. Among GHGs, the ones containing carbon (C) play a significant role as the dominant element in this heat-retention process10,11. While GHGs-generating sources— such as fossil fuel combustion, cement production, and other industrial processes—are the largest, biosphere sources also play a significant role12. Land biogenic GHG fluxes arise from natural carbon reservoirs and are shaped by both natural variability and human-induced disturbances13,14. A vital component of the biosphere is peatlands—terrestrial wetland ecosystems where waterlogged conditions inhibit the complete decomposition of plant material, resulting in the gradual accumulation of peat over time15,16. Peatlands cover less than 3% of the Earth’s surface but store more C than the combined biomass of the world’s tropical rainforests and over half of all the C currently in the atmosphere17–21. To affirm the great importance of peatlands, it suffices to say that northern boreal and temperate peatlands are the largest natural terrestrial carbon storage on the earth22,23. Although the ability of peatlands to function as a C sink rather than a source shows their pivot role in mitigating GHGs24–27 the degradation of peatlands transforms this role into a dramatically harmful one, resulting in a substantial increase in C emissions28,29. Peatland degradation leads to a reduction in peat volume, which manifests as subsidence— the downward movement of the ground surface. Therefore, given the direct relationship between degradation and C loss, as well as the link between degradation and subsidence, the subsidence rate can serve as a proxy for estimating C emission from peatlands30,31. Using subsidence as a proxy requires a clear understanding of its driving factors in peatlands32,33. Subsidence indicators analyzed in a seasonal pattern can be used in quantifying bog breathing process. Subsidence is primarily driven by oxidation and shrinkage34–36. Oxidation involves the decomposition of organic matter in 1Centre for Climate Research, Warsaw University of Life Sciences-SGGW, Nowoursynowska 166, Warsaw 02-787, Poland. 2Faculty of Civil Engineering and Environmental Sciences, Białystok University of Technology, ul. Wiejska 45A, Białystok 15-351, Poland. email: [email protected] OPEN Scientific Reports | (2025) 15:34175 1 | https://doi.org/10.1038/s41598-025-15293-1 www.nature.com/scientificreports the peat, releasing C into the atmosphere37,38. Shrinkage, on the other hand, is associated with the loss of water content within the peat, which translates to compaction39,40. Several studies have utilized peat subsidence as a proxy for estimating C emissions. This estimation requires three key components: the oxidation component (α), peat properties (Cv), and the subsidence rate (Δhs). Since the amount of C emitted from degraded peatlands is caused by oxidation processes, distinguishing it from the shrinkage component is essential, and this is where α comes into play. Sari, et al.41 Khodaei, et al.42 Hayati, et al.43 Erkens, et al.44 Zhou, et al.45 Hooijer, et al.46 Mos, et al.47 Anshari, et al.48 Couwenberg, et al.49 and Wösten, et al.50 have estimated CO2 emissions based on the assumption that oxidation contributes a substantial portion of subsidence, attributing approximately 60–100% (α = 0.6–1) of the total subsidence51,52. On the other hand, Grønlund, et al.53 reported that in Norwegian peatlands, oxidation accounted for 30–50% of total subsidence, which aligns with findings from Dutch peatlands, where oxidation contributions ranged from 35 to 50%, as noted by Schothorst54. These values show the lack of transparency in determining oxidation components, confirming the urgent need to adopt a unified approach. Therefore, one of the primary goals of this study is to propose this unified approach inspired by a global meta-analysis research conducted by Ma, et al.55. The bulk density (BD) and soil organic carbon (SOC) content—the property of peat (Cv)—are the other important parameters in estimating C loss from subsidence rates. These parameters are typically obtained through laboratory analysis of soil samples; however, the challenges of obtaining accurate field measurements have often led to their being overlooked. As a result, many studies commonly use constant values of 80kg/m³ for BD and 55% for SOC. Agus, et al.49 express that the BD and SOC range 20–300kg/m3 and 18–58%, which implies ranging from 3.6 to 174kg-C/m3 for Cv. So, this study also aims to improve these parameters by utilizing the dataset provided by Loisel, et al.56 for northern boreal and temperate peatlands, and the one introduced by Agus, et al.51 for tropical peatlands. Recent advancements have made it possible to estimate peat subsidence rates (Δhs) using remote sensing (RS), particularly the Interferometric Synthetic Aperture Radar (InSAR) technique57–59. Ghezelayagh, et al.57 demonstrated the use of InSAR for peatland ecosystems, employing Sentinel-1 satellite processed via Alaska Satellite Facility (ASF) OnDemand InSAR cloud computing products. This approach includes systematic error reduction to improve interferogram quality, accounting for spatial and temporal baseline, atmospheric condition, and ground cover. After proposing a remote-sensing-based framework, this study evaluates C emissions by applying it to the Biebrza Valley in Poland and validating each parameter against field survey-based data. We believe this study represents a significant step toward developing a more accurate, fully remote-sensing-based framework for estimating carbon emissions from peatlands. With high accuracy, such a framework would facilitate timely and consistent C emission monitoring and provide cost-effective, spatially distributed, and easily accessible data. This enables us to do large-scale assessments of peatland carbon dynamics and emissions. Achieving this goal requires a collective effort and effective communication among experts in this domain who have access to more in-situ data. Results and discussions Framework description Figure 1 presents the framework for estimating C emissions based on the amount of C lost by employing peat property (Cv) and the loss of peat (α∆h s) —the proportion of subsidence is caused by the oxidation process rather than shrinkage. This approach requires four key parameters: subsidence rate (m/y), oxidation component (%), BD (kg/m3), and SOC content (g-C/kg) (Fig.1). In the following section, we detail the methods for acquiring these parameters without relying on field survey data, enabling us to achieve a fully RS-based approach to C emission estimation. The results for each parameter were validated against field survey-based data to assess their validity. The rate of subsidence (Δhs) Figure 2 shows the spatial distribution of vertical displacement rates across the study area obtained by implementing the remote-sensing-based indicator proposed by Ghezelayagh, et al.57. The results reveal that the range of annual vertical displacement in the case study is from − 4.5 (subsidence) to + 2cm/year (accumulation). The distribution histogram chart indicates that this case study’s mean annual subsidence rate is 1.4cm (an average of 0.014 ± 0.007m per year). However, for C emission estimates, only areas experiencing subsidence (negative vertical displacement) are considered, as emissions are directly linked to the oxidation of peat during subsidence. Therefore, this approach quantifies the subsidence rate rather than the total vertical displacement, as only subsidence contributes to carbon emissions. More detail is provided in supplementary material 1 and Table S1. Oxidation component (α) α represents the fraction of subsidence attributed to the oxidation of organic matter in peatlands, rather than processes like shrinkage or consolidation. Estimating α is crucial for determining the C emissions associated with peat subsidence. In this study, α was acquired by the following equations derived from a global meta-analysis conducted by Ma, et al.55. Table1 is derived by manipulating equations from Ma et al.‘s study (see Methods). The results (Fig.3) indicate that the oxidation component varies spatially across the study area due to differences in land use. On average, α was 34.46% (± 19), meaning approximately 35% of the subsidence is attributable to oxidation. This aligns with findings from previous studies in Dutch and Norwegian peatlands, which reported α values ranging from 0.3 to 0.5 (Schothorst55 and Grønlund, et al.53). The spatial distribution of α shows higher values in areas with agricultural land use (including permanent meadows and pastures), indicating that peat oxidation is more pronounced under such conditions. Conversely, Scientific Reports | (2025) 15:34175 2 | https://doi.org/10.1038/s41598-025-15293-1 www.nature.com/scientificreports/ lower α values are observed in grassland (including extensively used meadows), areas with higher moisture retention or minimal anthropogenic interference. Peat property (BD and SOC) BD and SOC content directly determine the amount of carbon stored in peat soils and the carbon released during oxidation. In this approach for northern boreal and temperate peatlands, BD and SOC are acquired based on an in-suit dataset introduced by Loisel, et al.56. The dataset provides a table in which the location of the interested areas can acquire these parameters (see methods). According to Table3 and taking into account the location of Biebrza in continental Europe, the BD value in the table is 0.120 ± 0.139g/m3, ~ 120kg/m³ (obtained from 410 field measurements), while SOC content is 38.9 ± 1.3% (from 60 sites), which means 389g-C/kg. These results show that the value of 46.7kg/m3 for Cv aligns with values typically reported for temperate peatlands, ranging from 46 to 82kg/m³51. To ensure accuracy, the estimated BD and SOC values were validated against field survey data collected from the case study (Fig.4). To assess the deviation in SOC estimates, we first calculated the difference between their central values (389 ± 13 and 377.5 ± 85.2g-C/kg), yielding a difference of 11.5. Next, we determined the combined standard deviation by taking the square root of the sum of the squares of the individual standard deviations (13 and 85.2). By normalizing the difference between the central values using the combined standard deviation, we obtained a normalized difference of approximately 0.133. By adopting a similar approach for BD (133.3 ± 57 and 120 ± 139kg/m³), this value was calculated at 0.0865. Additionally, to be more confident in assessing BD deviation, we also used in-situ data reported by Gnatowski, et al.60 from the Biebrza River Valley, based on 87 samples, which indicated an average bulk density of 132 ± 38kg/m³. The normalized difference for BD was approximately 0.125. Fig. 1. Remote sensing (RS)-based framework for estimating C emission (CO2−eq/ha/year). The estimation is based on the peat loss (αΔhs) and peat properties (Cv), where Δhs represents the subsidence rate, α the oxidation component, ρpeat bulk density and fOC soil organic carbon content. Scientific Reports | (2025) 15:34175 3 | https://doi.org/10.1038/s41598-025-15293-1 www.nature.com/scientificreports/ The slight difference between the central values relative to their combined uncertainties, 0.133 for SOC and 0.0865 for BD, indicates that the estimated and measured values are in good agreement. Estimation of C emission The C emission rate is expressed in tons of CO2-eq per hectare per year. This metric provides a unified approach by converting all carbon emissions to CO₂, allowing for better comparison with other studies, as most research in this field reports emissions in terms of CO₂. Figure5 illustrates the C emission rate and its distribution across the case study. The histogram shows that the C emission rate in Biebrza Valley is 7.49 ± 3.6 ton-CO2-eq/ha/year. Peatland type Equations Northern boreal Peatlands Agriculture α=2.15 + 12.05ln(− 0 . 58 √ ∆h s 13.95 ) Forestry α=2.15 + 12.05ln(− 0 . 83 √ ∆h s 5.36 ) Grassland α=2.15 + 12.05ln(− 0 . 36 √ ∆h s 5.55 ) Tropical Peatlands Agriculture α=2.15 + 14.36ln(−0.37 √ ∆h s 6.63 ) Forestry Table 1. The equations for determining oxidation components in Northern peatlands using rate of subsidence (cm/year). Fig. 2. Spatial distribution of annual vertical displacement rates in the Biebrza Valley. Negative values indicate peat subsidence, while positive values represent peat accumulation. These rates were derived using the remotesensing-based indicator proposed by Ghezelayagh, et al.57. Scientific Reports | (2025) 15:34175 4 | https://doi.org/10.1038/s41598-025-15293-1 www.nature.com/scientificreports/ Moreover, the common approach for estimating CO2-eq emissions from peatlands using subsidence as a proxy assumes constant values for BD (80kg/m³) and SOC content (550 g-C/kg), with an oxidation component range of 30–100%. Using this approach, we estimated CO2-eq emission rates of 14.49 tons/ha/year, assuming a 60% oxidation rate commonly used in previous studies (see introduction). Several studies have employed this approach: Khodaei, et al.42 estimated C emissions in Sweden at approximately 6.7 ton-CO2−eq/ha/year based on a 0.9cm annual subsidence rate; Leifeld, et al.61 reported a range of 9.2–20.2 ton-CO2−eq/ha/year in exchange for subsidence rates of 0.8–1.6cm/year in European’s temperate fens. However, the estimation derived from the proposed framework is more consistent than the common approach with the findings of Wilson, et al.62 who reported C emissions from peatlands in Canada, the Republic of Ireland (ROI), the United Kingdom (UK), and Fennoscandia to be approximately 6.2 (± 0.47) ton-CO2−eq/ ha/year for industrial extraction sites and 6 (± 0.44) ton-CO2−eq/ha/year for domestic extraction sites in the ROI and UK. These values are notably lower than the IPCC63 Tier 1 default emission factor of 10.3 (± 1.7) tonCO2−eq/ha/year. Wilson et al.’s study highlights the necessity of refining regional emission factors to improve the accuracy of greenhouse gas inventories . Höper, et al.64 found that rates of carbon emissions from peatlands in Central European countries, including Poland, Germany, the Netherlands, and Sweden, range from 15 to 17 tonCO2−eq/ha/year. Boreal fens under grass and barley cultivation in Finland release around 22 ton-CO2−eq/ha/year. Depending on drainage intensity and management, forestry-related emissions range from net sequestration of −8 ton-CO2−eq/ha/year to emissions of 10 ton-CO2−eq/ha/year. Byrne, et al.65 report that high-emission regions in European peatlands—the Southeast Mediterranean, Germany, and the Netherlands—exhibit emissions exceeding 12.8 ton-CO2-eq/ha/year, primarily due to intensive agricultural use. Meanwhile, intact peatlands in Finland and the UK emit less than 4.04 ton-CO2-eq/ha/y. Significant differences in estimates occurred between the RS approach and the common approach: the RS-based approach calculates an annual emission of 7.49 tons CO2-eq per hectare, while the common approach estimates 14.49 ton-CO2-eq/ha/year (Table2; Fig.6). Fig. 3. Spatial distribution of the oxidation component (%) across the Biebrza Valley, showing the proportion of peat subsidence attributed to oxidation processes. Land use classes include agricultural land, which consists of permanent meadows and pastures; grassland, which refers to extensively used meadows; and forest. Scientific Reports | (2025) 15:34175 5 | https://doi.org/10.1038/s41598-025-15293-1 www.nature.com/scientificreports/ Fig. 5. The distribution of the C emission rate in the Biebrza Valley (ton-CO2-eq/ha/year) estimated through rate of subsidence (m/year), oxidation component (%), Bulk density (kg/m3), and Soil organic carbon content (g-C/kg). Fig. 4. Field-measured data for BD and SOC content in the Biebrza Valley. The data were utilized to validate the BD and SOC estimates derived from the RS framework approach. Scientific Reports | (2025) 15:34175 6 | https://doi.org/10.1038/s41598-025-15293-1 www.nature.com/scientificreports/ Conclusions This study lays the foundation for developing a comprehensive, fully remote sensing-based approach to improve the accuracy of peatland C emission estimates using peat subsidence as a proxy. By utilizing subsidence rates, peat properties, and the oxidation component, this method makes C emission estimation more accessible to researchers and managers. Although the presented methodology requires calibration elsewhere, the computational tool we presented and calibrated to the actual conditions of an extensive peatland seems ready for application in every possible peatland in the world, opening up the field for large-scale analysis in areas where field studies are significantly hampered (e.g., the vast peatlands of Canada and the West Siberian Plain). In these places, the uncertainty of the developed method gives way to the scale of the carbon emission and accumulation phenomenon being assessed, providing the appropriate order of magnitude of the assessed process. Refining this algorithm requires collaboration among experts with access to in-situ data and specialized knowledge of each parameter. A critical step in this improvement is evaluating the accuracy of acquisition methods for oxidation Fig. 6. The comparison of C emission estimation between the RS-based and the common approach. The RS approach Common approach C emission rate (ton-CO2-eq/ha/year) 7.49 14.49 Table 2. Summary of Estimation of the annual C emission (ton-CO2−eq/ha/year) obtained from different approaches. Scientific Reports | (2025) 15:34175 7 | https://doi.org/10.1038/s41598-025-15293-1 www.nature.com/scientificreports/ components, BD, and SOC. The primary objective of this study was, however, not only to propose an approach for estimating C emissions using subsidence as a proxy but also to assess the reliability of key parameters. The results indicate that the estimated C emission rate for the case study was 7.49 ton-CO2-eq/ha/year using the proposed framework, compared to 14.49 ton-CO2-eq/ha/year obtained from the conventional approach. The normalized differences of 0.133 for SOC and 0.0865 for BD further support the reliability of this method for assessing peat properties. Nevertheless, further research is needed to enhance the accuracy of C emission estimates derived from peatland subsidence. Materials and methods Study area Biebrza Valley, located in northeastern Poland (Fig.7), is an ecologically diverse region renowned for its glacialformed peat landscape, which is covered with a peat layer reaching a max 8m thickness and rich biodiversity66,67. The valley encompasses a variety of ecosystems, including peatlands, floodplains, and marshes, making it one of the most significant peatland complexes in Europe68 providing a critical habitat for numerous unique plant and animal species69. Recently, Biebrza Valley has faced significant anthropogenic pressures, primarily through land use changes and drainage measures intended to boost agricultural productivity. These interventions have led to subsidence phenomena, where the draining of peatlands has caused the ground to sink, presenting considerable environmental challenges70. The climatic conditions in Biebrza Valley are characterized by a mean annual temperature of around 6.6°C71 and a mean annual precipitation of 570mm (ranging from 470 to 730mm)72. Based on more than a decade of groundwater level measurements, a general decreasing trend of −0.21m has been observed73. In the Biebrza Basin, vary in thickness and stratigraphy depending on the type and extent of water supply. Peats in the 1–2m thickness range predominate, while the deepest thickness of the peat layer reaches over 6m. Fig. 7. Location of the Biebrza Valley, 53o12′49″N-53o44′45″N and 22o26′00″E-23o30′44″E, in northeastern Poland. Source of background map: Esri World Topographic Map (accessed on 20.06.2025) Scientific Reports | (2025) 15:34175 8 | https://doi.org/10.1038/s41598-025-15293-1 www.nature.com/scientificreports/ The relationship between C emission and peat subsidence The annual C emissions from peatland oxidation through a peatland subsidence proxy can be quantified using Eq.(1), expressed in units of tons of CO₂-equivalent per hectare per year. Any other C emissions through loss of peat through for example, erosion, are not part of this equation: C emission = α∆h s × ρ peat × fOC ×3 . 67 ×10−2 (1) where:C emission: The rate of Carbon loss (ton CO2-eq/ha/year);α∆hs: The rate of Peat loss, oxidation component of subsidence, (m⁄year); ρpeat: Bulk density (kg⁄m3 ); fOC: Soil organic carbon content (g-C/kg); 3.67: The conversion factor from C to CO2;10−2: The conversion factor from g/m2 to ton/ha Equation(1) is employed to estimate C emissions resulting from peat subsidence by incorporating the subsidence rate ( ∆h s ), the oxidation component ( α) , and the peat property (Cv)—including the BD (ρpeat) and SOC content (fOC). This study uses CO₂-equivalent (CO2-eq) to represent C emission because it provides a straightforward and comparable measure of greenhouse gas emissions. It is necessary to mention that the term CO2-eq is explicitly used to denote the mass conversion of elemental C to carbon dioxide (CO₂) based on molecular weight differences, applying a factor of 3.67. This factor accounts for the ratio of CO₂ (44g/ mol) to C (12g/mol), representing the complete oxidation of C to CO₂. In climate science, CO2-eq is often used to express the global warming potential (GWP) of various greenhouse gases relative to CO₂. However, in the context of this study, CO2-eq strictly refers to the mass-based conversion and should not be confused with GWP-based equivalence. This approach aligns with the fundamental principles of the POXAPS (Peat OXidation And Permanent Shrinkage) model, which estimates the contribution of oxidation and shrinkage to the overall subsidence31,43. The POXAPS model elucidates the relationship between the decomposition of organic matter and CO₂ emissions50,74. Previous studies have supported the use of this equation to estimate CO₂ emissions in both northern and tropical peatlands41–48. Acquisition of Peat Subsidence Rate (Δhs) This study employs the remote sensing-based approach introduced by Ghezelayagh, et al.57 to estimate peat surface vertical displacements. This method utilizes interferometric synthetic aperture radar (InSAR) techniques, leveraging Sentinel-1 satellite data processed through the ASF OnDemand InSAR cloud computing platform. To enhance the accuracy of interferometric measurements, the methodology incorporates a systematic error reduction framework that accounts for key factors, including satellite positioning, atmospheric conditions, vegetation cover, and temporal spacing between radar acquisitions. A detailed explanation of this approach is provided in Ghezelayagh, et al.57. Acquisition of peat properties (Cv) There are three approaches to assess peat properties (Cv) (BD and SOC) in this study. These three approaches provide different methods for obtaining peat properties, each with its own set of advantages and limitations. Field-based sampling offers high accuracy but is resource-intensive; the RS framework—the approach proposed in this study—provides a broad and efficient alternative but may require validation against ground data; the common approach offers simplicity and consistency based on established research but may not be accurate enough. In this study, we used the results obtained from the Field-based sampling approach as a criterion for validating two other approaches. • Constant value (a common approach in peatland studies). In peatland studies, a common approach is to use a constant value for BD and SOC content, typically held constant at 0.08g/cm³ (80kg/m³) and 55% (550 g-C/kg), respectively31,75,76. • RS framework approach. In northern boreal and temperate peatlands, an alternative approach for acquiring bulk density (BD) and soil organic carbon (SOC) data can be achieved by utilizing the comprehensive dataset compiled by Loisel, et al.56. This dataset provides BD and SOC values derived from an extensive collection of peat core samples, offering a robust and widely applicable reference for similar environmental studies. The dataset includes a synthesis of peatland soil properties constructed across 215 sites located north of 45°N. The values provided encompass different peatland types, including ombrotrophic bogs, minerotrophic fens, and permafrost-affected peatlands, making it a valuable resource for estimating soil properties in various case studies. BD and SOC values can be obtained by identifying the closest comparable peatland sites within the dataset (Table3) in terms of geographical location, climatic conditions, and peatland type. As shown in Table3, BD and SOC values can be extracted for the most relevant peat cores that match the locational, ecological, and hydrological conditions of the study area. These values were then applied to the case study region to estimate BD and SOC without needing direct field sampling. These values are also acquired from the study by Agus, et al.49 for tropical peatlands in which they used 1826 in-situ field sampling. Indeed, instead of conducting in-situ field sampling and laboratory analysis to determine BD and SOC, these values are extracted from the dataset developed by Loisel, et al.56 for northern boreal and temperate peatlands and Agus, et al.51 for tropical peatlands. Utilizing an existing dataset eliminates the need for extensive fieldwork and laboratory processing, making it an effective approach for studies constrained by logistical or financial limitations. By demonstrating the applicability of this dataset for BD and SOC estimation, it provides a replicable framework for its use in other regions where direct field data may be unavailable. It is necessary to mention that Scientific Reports | (2025) 15:34175 9 | https://doi.org/10.1038/s41598-025-15293-1 www.nature.com/scientificreports/