Changes in carbon stock in the Xing’an permafrost regions in Northeast China from the late 1980s to 2020

ABSTRACT

The degradation of Xing’an permafrost affects the stability of carbon pool and carbon emissions of the hemiboreal ecosystem in Northeast China. Due to the lack of long-term monitoring and detailed research, changes in carbon stock in terrestrial ecosystems in the Xing’an permafrost regions in Northeast China remain little known. In this study, we conducted more accurate simulations for the aboveground biomass (AGB), belowground biomass (BGB), dead organic carbon (DOC), and soil organic carbon (SOC) in the top 0–30 cm in permafrost regions in Northeast China using the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) model. The model was applied using multi-period land use/land cover (LULC) and carbon density data from the late 1980s, 2000, 2010, and 2020. In Northeast China, the permafrost extent shrank rapidly by 108,600 km² from 365,300 km² in the 1980s to 256,700 km² in the 2010s. The total carbon stock in permafrost regions in Northeast China was estimated at 4293.04 Tg C in the 1980s and at 4049.56 Tg C in the 2010s. Based on the permafrost extent of the 1980s, it was estimated that from the late 1980s to 2020, the LULC transformation resulted in the reduction of carbon stock by 33.53 Tg C, while the carbon fixation of vegetation growth increased by 1205.18 Tg C, i.e. a net carbon accumulation of 1149.15 Tg C. However, this estimate of the increasing carbon pool still awaits more systematic studies on the carbon budgets at larger soil depths of the top 2–3 m soils. This estimate can provide a timely preliminary estimate of the carbon pool as the baseline in the permafrost region of Northeast China for national and regional initiatives of carbon neutralization.

KEYWORDS:

• Climate Change
• Xing’an permafrost
• Carbon
• Integrated valuation of ecosystem service and tradeoffs (InVEST) model
• Boreal ecosystem

1. Introduction

As an important part of the global organic carbon (OC) budget, the OC pool in terrestrial ecosystems plays key roles in climate changes, such as feeding back to the global climate system by regulating the concentrations of some major atmospheric greenhouse gases (e.g. CO₂, CH₄, and N₂O) (Lal 2004; Stankovic et al. 2021). The carbon fixation capacity of different land use/land cover (LULC) varies significantly, and its change is the main driver for variations in OC in terrestrial ecosystems (Cantarello, Newton, and Hill 2011; Zhang et al. 2015; Stocker et al. 2017). In particular, as an important carbon sink in the global carbon cycle, the forest ecosystem stores over 80% of all terrestrial aboveground OC and over 70% of all soil organic carbon (SOC) (Jobbágy and Jackson 2000; Mu et al. 2013a). Changes in the forest areal extent and interannual variations in forest biomass dominate the long-term net carbon flux between terrestrial ecosystems and the atmosphere (Houghton 2005; Pugh et al. 2019). The biomass OC of the forest increases with the growing forest age, and the carbon sink function differs substantially among various forest types (Wei et al. 2013; Hu et al. 2015a; Hu et al. 2015b).

Permafrost is an important carbon reservoir with massive OC and high carbon density, and permafrost regions contain twice as much carbon as there is currently in the atmosphere (Zimov, Schuur, and Chapin 2006; Broderick et al. 2015; Ding et al. 2017; Mu et al. 2020). The effects of permafrost degradation on climate change depend mainly on SOC (Wu et al. 2021). In permafrost regions, the OC in the near-surface active layer accounts for 12% of all SOC, and the rest 88% is almost all found in the permafrost layer(s) (Zimov, Schuur, and Chapin 2006). Permafrost regions in the Northern Hemisphere contain a total carbon pool of 1100–1500 Pg C (Hugelius et al. 2014). With climate warming, permafrost has degraded extensively and rapidly (e.g. Li et al. 2022c), resulting in rapid and massive carbon losses (Zimov, Schuur, and Chapin 2006). About 5–15% of the permafrost carbon is immediately vulnerable to ground thawing (Schuur et al. 2015). The release of permafrost carbon alters not only the carbon source and sink functions of terrestrial ecosystems but also the carbon cycle in permafrost regions. This has led to a significant increase in atmospheric concentrations of CO₂, CH₄, N₂O, and air temperature (Harden et al. 2012; Ren et al. 2020; Jin and Ma 2021; Yang et al. 2021).

In a warming climate and increasing human activities, permafrost degradation has accelerated transformation in LULC patterns, such as the shrinkage of marshes due to permafrost degradation (Jin et al. 2007, 2008). This would result in modifications or shifts in the carbon and nitrogen cycling rates of terrestrial ecosystems (Jin et al. 2007; He et al. 2009; Mahowald et al. 2017; Turetsky et al. 2019; Wang et al. 2022b). Therefore, under a changing climate, systematic studies on the carbon cycling of terrestrial ecosystems in permafrost regions and the exchange processes of OC between permafrost and active layer can more accurately explain the impacting mechanisms of carbon sources and sinks (Ni et al. 2019; Dong et al. 2021). A recent synthesis based on extensive data has estimated that by 2100, the release of SOC to the atmosphere from the northern permafrost regions will be 12–113 Pg C (Koven et al. 2015). The results of model simulations under the Representative Concentration Pathway (RCP) 4.5 scenario indicate a loss of permafrost extent by 3–5 million km² between 2010 and 2299 (McGuire et al. 2018). For the RCP4.5 projection, the cumulative change in SOC varies between 66 Pg C loss to 70 Pg C gain in the northern permafrost regions; for the RCP4.5 projection, gains in vegetation carbon are largely responsible for the overall projected net gains (8–244 Pg C) in ecosystem carbon by 2299 (McGuire et al. 2018). The terrestrial carbon cycle is currently the least constrained component of the global carbon budget (Bloom et al. 2016). Large uncertainties stem from a poor understanding of plant carbon allocation, stocks, residence times, and carbon use efficiency (Bloom et al. 2016). The residence times of live biomass and dead OC (DOC) exhibit contrasting spatial distributive features of carbon density and biogeochemistry (Bloom et al. 2016). At present, the methods of field investigations, flux observations, remote sensing monitoring, and model simulations are mainly used to estimate regional carbon stocks (e.g. Chen et al. 2000; Yu et al. 2013; Zhang et al. 2013). Field surveys and flux observations are accurate on a single point or quadrangle scale but require tedious and costly fieldwork. With the rapid development of remote sensing technology, the key parameters of carbon storage simulation can be readily obtained, which helps model simulation become one of the mainstreams for estimating the carbon storage of terrestrial ecosystems (Zhao et al. 2019). LULC can directly characterize the impact of human activities and global changes on ecosystems and is an important input parameter to simulate global climate effects and biogeochemical effects (Duarte et al. 2016; Liu et al. 2018; Mohajane et al. 2018). The Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) model can consider aboveground biomass (AGB), belowground biomass (BGB), DOC, and SOC of different LULC types. Therefore, it can more reliably estimate carbon budgets and processes, which have been widely applied in regional OC simulations (e.g. Xiang et al. 2018; Ma et al. 2019; Babbar et al. 2021).

Northeast China is the second-largest permafrost region in China, and the northern part of this region also has one of the major forested areas in China (Zhou et al. 2000). The symbiotic mechanisms among permafrost, forest, wetland, and grassland ecosystems results in the storage of large amounts of OC in permafrost regions (Jin et al. 2008; Wei et al. 2011; Hugelius et al. 2014). Under the combined influences of climate warming and human activities, permafrost degradation has been extensive and intensive in Northeast China since the 1980s (Jin et al. 2000, 2007; Zhang et al. 2021; Șerban et al. 2021; Li et al. 2022b). The degradation of permafrost results in the degeneration of the hemiboreal forest ecosystem, the northward and upward shifts of forest lines, the shrinkage of wetlands, and the evolution from wetland ecosystem to grassland ecosystem (Jin et al. 2008; Wei et al. 2011; Chen, Zang, and Sun 2020). Furthermore, permafrost degradation affects the carbon pool, OC release, and carbon cycling processes of the northern and upland ecosystems (Duan et al. 2020; Mu et al. 2017; Jin and Ma 2021). However, how does the carbon sequestration potential of each ecosystem change in the Xing’an permafrost regions of Northeast China? Additionally, soil organic carbon content, carbon pool size and thermal stability, carbon emission and carbon fixation rate in permafrost areas are poorly understood. Therefore, this paper improved the carbon storage estimation module of the InVEST model by introducing data on forest age and vegetation type. This study aims to estimate the ecosystem carbon storage and analyze the carbon storage changes in the Xing’an permafrost regions in Northeast China from the late 1980s to 2020. This study is of great significance for understanding the roles of long-term trends in carbon sources/sinks and carbon cycles and their responses to climate change in permafrost regions. The study results can timely provide reference data for carbon neutralization initiatives in China and sustainable development of ecological environments in northern permafrost regions.

2. Materials and methods

The carbon density of different LULC types in Northeast China was obtained through literature review, vegetation map and forest age map, and the relationship between forest age and carbon density was established for varied forest types. Based on LULC data from the late 1980s to 2020 and carbon density data of different LULC types, the AGB, BGB, DOC, and SOC (0–30 cm) in Northeast China were estimated through the improved carbon storage estimation module of the InVEST model. Using the distribution of permafrost from the 1980s to the 2010s and three different scenarios (actual: changing LULC and changing carbon density, scenario 1: changing LULC and constant carbon density, and scenario 2: changing carbon density and constant LULC), changes in carbon stock in the Xing’an permafrost regions in Northeast China from the late 1980s to 2020 were evaluated. Based on three different scenarios, changes in carbon stock in the Xing’an permafrost regions in Northeast China from the late 1980s to 2020 were evaluated (Figure 1).

Figure 1. The workflow of studying the change of carbon storage in Xing’an permafrost regions in Northeast China.

2.1. Study area descriptions

The northern part of Northeast China is one of the major permafrost regions in China. It is in the sensitive and fragile ecotone on the southern margin of the discontinuous and sporadic/patchy latitudinal permafrost zones on the East Asian continent (Figure 2 (a)) (Jin et al. 2007). Under the significant influences of changing hydroclimatic and other local geo-environmental factors, the Xing’an permafrost was formed in Northeast China during the past cold climatic conditions. Overall, it has been degrading since the Last Glaciation Maximum (21 ± 2 ka BP) (Jin et al. 2019). The unstable and warm (> −1°C) Xing’an permafrost is dominated (driven, changed, or protected) by the hemiboreal ecosystem, in contrast to the stable and cold climate-driven permafrost at very high latitudes or elevations. The zones of predominantly continuous (70–80%) and island discontinuous (30–70%) latitudinal permafrost were mainly found in the northern Da Xing’an Mountains (Figure 2 (b)) (Zhou et al. 2000). Island (30–70%) and sporadic island (5–30%) latitudinal permafrost were mainly distributed in the Xiao Xing’an Mountains and on the northern Hulun Buir and Songhua-Nen river plains (Figure 2 (b)) (Zhou et al. 2000). Mountain permafrost also occurred at elevations of above 1500–1700 m asl in the Changbai and southern Da Xinganling mountains, such as the Huanggangliang Mountains, to the south of the southern limit of latitudinal permafrost (Figure 2 (b)) (Zhou et al. 2000; Jin et al. 2007). Boreal forest and wetland ecosystems prevail in the Xing’an permafrost regions in Northeast China. Among them, the vegetation of the Da Xing’anling Mountains belongs to the cold temperate coniferous forest (hemiboreal forest) dominated by the Xing’an larch (Larix gmelinii). The vegetation in the Xiao Xing’an Mountains belongs to the coniferous and broadleaved mixed forest in the North Temperate Zone, in which broadleaved Korean pine (Pinus koraiensis) forest dominates (Hu et al. 2015b). Affected by climate change, forest fires, and engineering construction and operation, the active layer deepens, the permafrost table lowers, ground temperature rises, and talik thickens, expands, and/or penetrates through the thin and warm permafrost layer. The Xing’an permafrost has been degraded extensively in the past few decades (e.g. Jin et al. 2007; He et al. 2009, 2021; Li et al. 2021a; Li et al. 2021b; Șerban et al. 2021; Zhang et al. 2021). Permafrost degradation may have accelerated the northward movement of the hemiboreal forest belt dominated by the Xing’an larch (Larix gmelinii) and may have facilitated the shrinkage of boreal wetlands and peatlands (Jin et al. 2007, 2008; Chen, Zang, and Sun 2020). As a result, greenhouse gas emissions increase, affecting the landscape ecology, carbon and nitrogen cycles, and water and ecological security in cold regions (Jin et al. 2008; He et al. 2009, 2021; Chen, Zang, and Sun 2020).

Figure 2. Map of permafrost types in Northeast China and its location in permafrost regions in the Northern Hemisphere.

Notes: (a) Distribution of permafrost with 1 km resolution for the period of 2000–2016 in the Northern Hemisphere (modified from Ran et al. 2021); (b) Distribution of permafrost types with 100 m resolution for the 2010s in Northeast China (modified from Wang et al. 2022a).

2.2. Data

2.2.1. Land use/land cover (LULC)

The Chinese Academy of Sciences established an 1:100,000 scale remote-sensing monitoring database for LULC in China (late 1980s, 2000, 2010, and 2020) based on Landsat thematic mapper (TM), enhanced thematic mapper (ETM), and operational land imager (OLI) remote-sensing images using human-computer interactive visual interpretation. The overall accuracy of the data is better than 85% (Xu 2014). The LULC data can be divided into six primary categories and 25 secondary categories, including cultivated land (dry land and paddy land), forestland (forestland, shrub land, sparse woods, and other woods), grassland (high coverage grassland, moderate coverage grassland, and low coverage grassland), waterbodies (rivers and men-made ditches, lakes, reservoirs and pit-pone, and bottomlands, and others), construction land (urban land, rural area, and other built land), and unused land (desert, the Gobi, saline and halic soils, wetlands, bare land, bare rock, and other lands). The LULC data used in this study are the raster data with a spatial resolution of 100 m. The dataset was downloaded from the Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences (http://www.resdc.cn).

2.2.2. Other data

The data of forest age use the spatial distribution map of forest age in China in 2001 (Wang et al. 2011; Zhou et al. 2016). The data were obtained from the fourth forest inventory of different provinces and cities in China from 1989 to 1993 and were rasterized by statistical method with a spatial resolution of 1 km. Based on the forest age data of 2001, the forest age distribution of other years was calculated by adding or subtracting time.

The data for vegetation were obtained from the 1:1,000,000 vector maps of vegetation distribution in China, which was compiled by Chinese scientists based on rich survey data and aerial remote sensing images (Zhang 2007). The vegetation map showed in detail the distribution and geographical patterns of vegetation-type groups, vegetation types, vegetation groups and subgroups, and dominant species of plant communities in China in 2001. The dataset was downloaded from the National Tibetan Plateau Data Center (http://data.tpdc.ac.cn).

Permafrost distribution data are based on long-term temperature observations from meteorological stations, combined with a digital elevation model (DEM), and generated slope angles and aspects. The method of ordinary least squares (OLS) was used to build overall optimized 3-D zonation (OO3DZ) models of mean annual air temperature (MAAT) and topographic data in Northeast China for MAAT simulation. The Kriging method was combined to minimize the MAAT residuals. On this basis, the spatial distribution map of MAAT with high accuracy (R > 0.97; Root mean square error (RMSE) < 0.7°C) in Northeast China was calculated (Wang et al. 2022a). The permafrost zone classification system in Northeast China established by Zhou et al. (2000), based on rich field survey data and MAAT, is used to divide permafrost into the zones of discontinuous, island, and sporadic permafrost. Permafrost distribution data from the four periods of the 1980s, 1990s, 2000s, and 2010s in Northeast China were used. With a high accuracy and a spatial resolution of 100 m, the data can more accurately reflect the permafrost distribution in Northeast China and its changes over the last four decades.

2.3. Methods

2.3.1. Parameters used in the InVEST model

By introducing the vegetation map, forest age data, and carbon density values from the literature, we obtained carbon density tables of AGB, BGB, DOC, and SOC (0–30 cm) of different LULC types in Northeast China (Table 1). According to the LULC and vegetation maps, the coniferous forests in Northeast China are mainly dominated by Larix gmelinii and Pinus sylvestris var. mongolica Litv. Broad-leaved forests are mainly dominated by Betula platyphylla and Quercus mongolica. Coniferous and broadleaf mixed forests are mainly by Pinus koraiensis and Quercus mongolica. Field investigation data of carbon density of different forest types and age groups in the literature were used. The regression relationship between forest age and carbon density values (AGB, BGB, DOC, and SOC (0–30 cm)) of coniferous, broadleaf, and coniferous-broadleaf mixed forests was established. The carbon densities of other LULCs were assigned according to the results of carbon density reviewed from the literature of the study area.

Table 1. Carbon density of each land use/land cover (LULC) in the InVEST model.

LULC		Aboveground biomass density (Mg/hm^2)	Belowground living density (Mg/hm^2)	Dead organic carbon density (Mg/hm^2)	Soil organic carbon density (0–30 cm in depth) (Mg/hm^2)	Data sources
Forestland	Coniferous forest	$C_{i\mathrm{\_}above}$=0.7560×Age–11.9130 (R^2=0.99)	$C_{i\mathrm{\_}below}$=0.0686×Age+0.0135 (R^2=0.99)	$C_{i\mathrm{\_}dead}$=0.0358×Age+2.6421 (R^2=0.95)	$C_{i\mathrm{\_}\mathrm{s}\mathrm{o}\mathrm{i}\mathrm{l}}$=1.2994×Age+44.9140 (R^2=0.82)	(Hu et al. 2015b; Ge 2018)
	Broad-leaved forest	$C_{i\mathrm{\_}above}$=0.9227×Age–14.7140 (R^2=0.96)	$C_{i\mathrm{\_}below}$=0.0592×Age–0.2562 (R^2=0.97)	$C_{i\mathrm{\_}dead}$=0.0318×Age+0.1675 (R^2=0.94)	$C_{i\mathrm{\_}\mathrm{s}\mathrm{o}\mathrm{i}\mathrm{l}}$=2.8733×Age–13.229 (R^2=0.87)	(Hu et al. 2015b; Wei and Man 2019)
	Coniferous and broad-leaved mixed forest	$C_{i\mathrm{\_}above}$=0.9246×Age–3.4534 (R^2=1.00)	$C_{i\mathrm{\_}below}$=0.0338×Age+0.2258 (R^2=1.00)	$C_{i\mathrm{\_}dead}$=0.0374×Age+2.8590 (R^2=0.56)	$C_{i\mathrm{\_}\mathrm{s}\mathrm{o}\mathrm{i}\mathrm{l}}$=1.1832×Age+76.3250 (R^2=0.43)	(Hu et al. 2015b; Ge 2018; Wei and Man 2019)
	Shrubs	9.37	4.69	2.23	78.04	(Wang 2016; Liu et al. 2019)
Grassland		0.52	4.62	0.00	60.80	(Liu et al. 2019; Xin et al. 2020)
Cultivated land		7.40	0.70	0.00	49.33	(Fang, Liu, and Xu 1996; Liu et al. 2019)
Wetlands		8.65	1.70	1.65	271.15	(Mu et al. 2013b)
Construction land		0.00	0.00	0.00	0.00	(Wang 2016)
Unused land		0.10	0.00	0.00	9.60	(Fang, Liu, and Xu 1996)
Water bodies		0.00	0.00	0.00	0.00	(Babbar et al. 2021)

$C_{i\mathrm{\_}above}$, $C_{i\mathrm{\_}below}$, $C_{i\mathrm{\_}dead}$ and $C_{i\mathrm{\_}soil}$ are carbon densities of aboveground biomass, belowground biomass, dead organic carbon, and soil organic carbon, respectively, at depths of 0–30 cm. The age is the average forest age.

2.3.2. Total carbon estimation model

The InVEST model was jointly developed by Stanford University, the Nature Conservancy, and the World Wildlife Fund. It can realize the spatial quantitative evaluation of ecosystem service values. The carbon storage module in the InVEST model uses carbon density and LULC as input data to estimate the total carbon of terrestrial ecosystems. Four carbon pools (AGB, BGB, DOC, and SOC (0–30 cm)) were used to estimate the total carbon in the current landscape or the change in total carbon over a given time period (Sharp et al. 2020). The regional total carbon was calculated as Eq.1

(1) $C_{Total}={\sum }_{i=1}^n{A}_i\times \left(C_{i\mathrm{\_}above}+C_{i\mathrm{\_}below}+C_{i\mathrm{\_}dead}+C_{i\mathrm{\_}soil}\right)$(1)

where $C_{Total}$ is the total carbon in the study region; $A_i$ is the areal extent of LULC type i, n is the total number of LULC types, and $C_{i\mathrm{\_}above}$, $C_{i\mathrm{\_}below}$, $C_{i\mathrm{\_}dead}$ and $C_{i\mathrm{\_}soil}$ are carbon densities of AGB, BGB, DOC, and SOC of LULC type i, respectively.

2.3.3. Calculation method for the contribution ratio of LULC transformation and vegetation carbon fixation to total carbon stock

To quantitatively evaluate changes in ecosystem total carbon stock caused by the LULC transformation and vegetation carbon fixation, two scenarios were designed to drive the InVEST model for estimating the change in ecosystem carbon stock. The Scenario 1 of “Changing LULC and constant carbon density” was used to evaluate changes in the total carbon stock caused by changes in the LULC transformation. The Scenario 2 of “Changing carbon density and constant LULC” was used to estimate the change of total carbon caused by vegetation growth.

Furthermore, the contribution ratio is used to calculate the contribution percentage of LULC transformation and vegetation growth to regional total carbon stock. The specific algorithm of the contribution ratio is expressed in Eqs. 2 and 3 (Yang et al. 2021):

(2) $R_l=\frac{\mathrm{\Delta }l}{\mathrm{\Delta }\mathrm{P}+\mathrm{\Delta }l}\times 100\mathrm{\%}$(2)

(3) $R_p=\frac{\mathrm{\Delta }p}{\mathrm{\Delta }\mathrm{P}+\mathrm{\Delta }l}\times 100\mathrm{\%}$(3)

where R_l is the contribution ratio of the LULC transformation to the total carbon stock. R_p is the contribution ratio of vegetation growth to total carbon stock. The △l is the change in total carbon stock under Scenario 1 of “Changing LULC and constant carbon density.” The △p is the change in total carbon stock under Scenario 2 of “Changing carbon density and constant LULC.”

3. Results

3.1. Changes in permafrost extents and zones in Northeast China

From the area transfer matrix of permafrost types, the areal extent of Xing’an permafrost in Northeast China changed significantly from the 1980s to the 2010s (Figure 3). In the 1980s, the areal extents of discontinuous, island, and sporadic permafrost zones were 3.85 × 10⁴, 8.30 × 10⁴, and 24.39 × 10⁴ km², respectively. In the 1990s, discontinuous, island, and sporadic permafrost all transformed into permafrost types of lower areal continuity, such as the transformation from discontinuous to island permafrost and island to sporadic permafrost. Regarding the areal extent, about 27% of discontinuous permafrost was converted to island permafrost. The changes in the areal extent of sporadic permafrost were the greatest, with 7.03 × 10⁴ km² (29%) of sporadic permafrost turned into seasonal frost. Similarly, the trend from the 1990s to the 2000s was consistent with that from the 1980s to the 1990s. The most obvious changes in areal extent occurred in regions of discontinuous and sporadic permafrost. From the 2000s to 2010s, a small amount of seasonal frost (0.98 × 10⁴ km²) was transformed into sporadic permafrost. In the 2010s, the areal extents of discontinuous, island, and sporadic permafrost were 0.86 × 10⁴, 7.42 × 10⁴ and 17.39 × 10⁴ km², respectively. Overall, from the 1980s to the 2010s, the Xing’an permafrost degraded substantially. The areal extents of discontinuous, island, and sporadic permafrost in Northeast China were reduced by 3.00 × 10⁴, 3.87 × 10⁴, and 10.86 × 10⁴ km², respectively (Table 2).

Figure 3. Transformation of permafrost types during the four periods of the 1980s–1990s, 1990s–2000s, 2000s–2010s and 1980s–2010s.

Notes: Pf stands for permafrost; Discont, discontinuous; Sporad, sporadic.

Table 2. Frozen ground zone matrix during the four periods of the 1980s–1990s, 1990s–2000s, 2000s–2010s, and 1980s–2010s (areal extent of frozen ground (AEFG) in 10⁴ km²).

AEFG	Pf Zones	Discont	Island	Sporad	Seasonal frost	Total
1990s	1980s
	Discontinuous	2.81	0.01	0.00	0.00	2.82
	Island	1.04	6.46	0.02	0.00	7.52
	Sporadic	0.00	1.83	17.34	0.00	19.17
	Seasonal frost	0.00	0.00	7.03	0.00	7.03
	Total	3.85	8.30	24.39	0.00	36.53
2000s	1990s
	Discontinuous	1.83	0.00	0.00	0.00	1.83
	Island	0.98	6.63	0.00	0.00	7.61
	Sporadic	0.00	0.89	15.08	0.00	15.97
	Seasonal frost	0.00	0.00	4.09	7.02	11.11
	Total	2.81	7.52	19.17	7.02	36.53
2010s	2000s
	Discontinuous	0.85	0.00	0.00	0.00	0.85
	Island	0.98	6.41	0.02	0.00	7.41
	Sporadic	0.00	1.19	15.22	0.98	17.39
	Seasonal frost	0.00	0.00	0.73	10.13	10.86
	Total	1.83	7.60	15.97	11.11	36.53
2010s	1980s
	Discontinuous	0.86	0.00	0.00	0.00	0.86
	Island	3.00	4.42	0.00	0.00	7.42
	Sporadic	0.00	3.87	13.52	0.00	17.39
	Seasonal frost	0.00	0.00	10.86	0.00	10.86
	Total	3.86	8.29	24.38	0.00	36.53

3.2. Spatiotemporal variations in total carbon stock in permafrost regions

Based on the carbon module of the improved InVEST model, the total carbon stock in Northeast China was estimated for the four time slots of the late 1980s, 2000, 2010, and 2020. The total carbon stock includes the AGB, BGB, DOC, and SOC (0–30 cm in depth). Figure 4 shows that the areas of high carbon stock in Northeast China are mainly distributed in forested areas in the Da and Xiao Xing’anling and Changbai mountains, as well as wetlands on the northern parts of the Songhua-Liao, Sanjiang (Three Rivers), and Hulun Buir plains. From the statistics of total carbon stock in permafrost regions from the 1980s to 2010s (Table 3), the total carbon pool in permafrost regions in Northeast China was 4293.04 Tg C in the 1980s. The total carbon stock in the zones of discontinuous, island, and sporadic permafrost accounted for 16.36%, 25.60%, and 58.04% of the total carbon pool in the permafrost zone, respectively. With the degradation and shrinking of the Xing’an permafrost, the total carbon pool in permafrost regions in Northeast China was reduced to 4049.56 Tg C in the 2010s, when the total carbon stocks in the zones of discontinuous, island, and sporadic permafrost accounted for 5.64%, 37.19%, and 57.17% of the total carbon pool in the permafrost zone, respectively.

Figure 4. Multi-period (late 1980s/2000/2010/2020) total carbon stock in Northeast China. Note: SLLP stands for the southern limit of latitudinal permafrost.

Table 3. Statistics of total carbon stock in permafrost zones in Northeast China from the 1980s to 2010s.

Periods	Discontinuous permafrost		Island permafrost		Sporadic permafrost		Total carbon (Tg C)
	Areal extent (10^4 km^2)	Total carbon (Tg C)	Areal extent (10^4 km^2)	Total carbon (Tg C)	Areal extent (10^4 km^2)	Total carbon (Tg C)
1980s	3.85	702.31	8.29	1098.91	24.39	2491.42	4292.64
1990s	2.82	601.83	7.52	1208.23	19.17	2219.70	4029.76
2000s	1.84	438.05	7.61	1378.87	15.98	1998.23	3815.15
2010s	0.86	228.42	7.41	1505.9	17.40	2315.25	4049.57

By comparing the average carbon density of different types of permafrost (Table 3), the average carbon density of permafrost regions in Northeast China was 117.52 MgC/hm² in the 1980s. The average carbon density in regions of discontinuous, island, and sporadic permafrost was 182.42, 132.56, and 102.17 MgC/hm², respectively. With elapsing time, the average carbon density of different permafrost types also gradually increased. The average carbon density of all permafrost regions in Northeast China was 157.75 MgC/hm² in the 2010s. These values included the average carbon density in zones of discontinuous, island, and sporadic permafrost at 265.62, 203.23, and 133.06 MgC/hm², respectively. Overall, the carbon density in permafrost regions of Northeast China was the largest (>50%) in the zone of sporadic permafrost. With the degradation of permafrost, the total carbon in permafrost regions gradually decreased during the period from the 1980s to the 2010s. The average carbon density of different permafrost types declines in the order of discontinuous permafrost, island permafrost, and sporadic permafrost. With elapsing time, the average carbon density of different permafrost types may gradually increase due to the possible expansion of sporadic permafrost zones.

3.3. Changes in carbon stock of different LULC types in permafrost regions

From the statistics and analysis of AGB, BGB, DOC, and SOC (0–30 cm in depth) and total carbon stock of different LULCs in permafrost regions in Northeast China (Figure 5), woodland dominates in permafrost regions. In the 1980s, the woodland area underlain by permafrost was 22.46 × 10⁴ km², accounting for 61.48% of the permafrost area. In the 1980s, the total carbon stock of woodland was 3174.94 Tg (AGB, 571.35 Tg, BGB, 96.37 Tg, DOC, 73.58 Tg, and SOC, 2433.64 Tg), accounting for 73.96% of the total carbon pool in permafrost regions. With the degradation of permafrost, the woodland area in the permafrost regions decreased to 17.07 × 10⁴ km² in the 2010s. In the 2010s, the carbon stock of woodland was 3393.25 Tg (AGB, 742.17 Tg, BGB, 100.14 Tg, DOC, 72.89 Tg, and SOC, 2478.05 Tg), accounting for 83.79% of the total carbon pool in permafrost regions. Although the woodland area was reduced, due to constantly growing vegetation, the capacity of woodland for carbon sequestration was also enhanced, leading to an increased total carbon amount in the 2010s. In the 1980s, the grassland area in the permafrost regions was 11.26 × 10⁴ km², accounting for 30.82% of the permafrost area. In the 1980s, the total carbon stock of grassland was 742.71 Tg (AGB, BGB, and SOC were 5.85, 52.05, and 684.54 Tg, respectively), accounting for 17.29% of the total carbon in permafrost regions. In the 2010s, the grassland area of the permafrost regions decreased to 7.06 × 10⁴ km², while the total carbon stock of grassland decreased to 465.28 Tg, accounting for 11.49% of the total carbon pool of permafrost regions. Despite the smaller areal extent of wetlands in permafrost regions, wetlands sequestered relatively more carbon per unit area. In the 1980s, the wetland area in the permafrost regions was 1.04 × 10⁴ km², accounting for 2.85% of the permafrost area. In the 1980s, the total carbon stock of wetlands was 293.58 Tg (AGB, 8.97 Tg, BGB, 1.76 Tg, DOC, 1.71 Tg, and SOC, 281.14 Tg), accounting for 6.84% of the total carbon in the permafrost area. In the 2010s, the wetland area in the permafrost region decreased to 0.51 × 10⁴ km², and total carbon decreased to 143.75 Tg, accounting for 3.60% of the total carbon in the permafrost area. The cultivated land area of the permafrost regions in the 1980s was 1.41 × 10⁴ km², and its total carbon was 81.11 Tg (AGB, BGB, and SOC were 10.45 Tg, 0.99 Tg, and 69.67 Tg, respectively), accounting for 1.89% of the total carbon storage in the permafrost regions. In the 2010s, the cultivated land area in the permafrost regions decreased to 0.78 × 10⁴ km², and total carbon decreased to 44.92 Tg, accounting for 1.11% of the total carbon in the permafrost regions. Unused land has low soil carbon density and less total carbon. In the 1980s and 2010s, the total carbon of unused land in permafrost regions was 1.03 and 0.424 Tg, respectively. The total carbon of construction land and waterbodies was assumed to be 0 Tg. In general, woodland has the largest total carbon storage, followed by grassland and wetland, and cultivated land, the smallest. For the total carbon stock, AGB and SOC contributed the most.

Figure 5. Statistical chart of different land use/land cover (LULC) types area and carbon in permafrost regions of Northeast China from the 1980s to 2010s. (a) Area of different LULC types; (b) Aboveground biomass (AGB) of different LULC types; (c) Belowground biomass (BGB) of different LULC types; (d) Dead organic carbon (DOC) of different LULC types; (e) Soil organic carbon (SOC) (0–30 cm) of different LULC types, and; (f) Total carbon of different LULC types.

3.4. Influencing factors for changing carbon stock in permafrost regions

Due to permafrost degradation and intense human activities, the forest line (timberline/treeline) in the northern part of Northeast China has been moving northward and the wetlands have shrunk, thus increasing the carbon release. Based on the area of the permafrost regions in the 1980s, changes in actual total carbon (AGB, BGB, DOC, SOC (0–30 cm in depth)) in Northeast China were analyzed for the four periods of the late 1980s, 2000, 2010, and 2020. Combined with the two scenarios, the influences of LULC transformation and vegetation carbon fixation on carbon stock were estimated. The total carbon increased from 4293.04 Tg C in the late 1980s to 5442.19 Tg C in 2020, with an increase of 1149.15 Tg C. The increase in total carbon mostly occurred in the period from the late 1980s to 2000, with an increase in carbon stock by 466.51 Tg C. From 2010 to 2020, the increase was the least (330.86 Tg C) (Table 4).

Table 4. Statistics of total carbon and contribution rates under different scenarios.

Periods	Total carbon (TgC)			Contribute ratio (%)
	Actual	Scenario 1	Scenario 2	Rl	Rp
Late 1980s	4293.04	4293.04	4293.04	/	/
2000	4759.55	4266.03	4788.41	/	/
2010	5111.33	4264.98	5143.32	/	/
2020	5442.19	4259.51	5498.22	/	/
Late 1980s–2000	466.51	−27.01	495.37	−5.77	105.77
2000–2010	351.78	−1.05	354.91	−0.30	100.30
2010–2020	330.86	−5.47	354.90	−1.57	101.57
Late 1980s–2020	1149.15	−33.53	1205.18	−2.86	102.86

Notes.

1) The total carbon for the late 1980s to 2020 was calculated based on permafrost regions in Northeast China in the 1980s.

2) Actual: ”Changing LULC and change carbon density” was used to estimate the total carbon stock.

3) Scenario 1: “Changing LULC and constant carbon density” was used to estimate the total carbon stock.

4) Scenario 2: “Changing carbon density and constant LULC” was used to estimate the total carbon stock.

5) R1 is the contribution ratio of LULC transformation to actual total carbon stock, and.

5) Rp is the contribution ratio of vegetation growth to actual total carbon stock.

Changes in total carbon (AGB, BGB, DOC, SOC (0–30 cm in depth)) under different scenarios were compared (Table 4). In Scenario 1 of “Changing LULC and constant carbon density,” total carbon stock continued to decline from the late 1980s to 2020. Total carbon decreased from 4293.04 Tg C (late 1980s) to 4259.21 Tg C (2020), with a carbon loss of 33.53 Tg C. The decrease in carbon was the most (27.01 Tg C) from the late 1980s to 2000. From 2000 to 2010, the decrease in carbon was the least (1.05 Tg C). In Scenario 2 of “Changing carbon density and constant LULC,” total carbon stock continued increasing from the late 1980s to 2020. Total carbon increased by 1205.18 Tg C from 4293.04 Tg C (late 1980s) to 5498.22 Tg C (2020). From the late 1980s to 2000, the increase in carbon was the largest (495.37 Tg C). The increase in carbon was similar to that in other years. The contribution ratio of total carbon in the two scenarios to actual total carbon also varied interannually. From the late 1980s to 2000, in Scenario 1, the contribution ratio of total carbon was −5.77%, and in Scenario 2, 105.77%. From 2000 to 2010, the contribution ratio of total carbon in Scenario 1 was −0.30%, while that in Scenario 2, 100.30%. From the late 1980s to 2020, the contribution ratio of total carbon in Scenario 1 was −2.86% and that in Scenario 2, 102.86%. In summary, by comparing the actual total carbon with the total carbon under the two scenarios, the LULC transformation resulted in the reduction of carbon stock. From the late 1980s to 2020, the loss of carbon storage was mainly concentrated in the conversion of forestland to grassland (33.92 Tg), forestland to cultivated land (17.26 Tg), forestland to waterbodies (9.39 Tg), wetlands to cultivated land (5.83 Tg), wetlands to grassland (5.62 Tg), and grassland to cultivated land (3.30 Tg). Additionally, the increase in carbon storage was mainly concentrated in the conversion of grassland to wetlands (14.01 Tg), grassland to forestland (10.97 Tg), forestland to wetlands (10.04 Tg), waterbodies to wetlands (4.21 Tg), and cultivated land to wetlands (3.59 Tg). Vegetation growth boosted carbon stocks in the Xing’an permafrost regions. The increase of vegetational carbon fixation was higher than the carbon release due to LULC transformation. As a result, the total carbon pool in permafrost regions in Northeast China rose by 1149.15 Tg C from the late 1980s to 2020.

4. Discussions

4.1. Reliability analysis of carbon simulation

The InVEST model can estimate the total carbon pool from four carbon stocks of AGB, BGB, DOC, and SOC at depths of 0–30 cm in different types of LULC (He et al. 2016). The forestland dominates in the LULC types in permafrost regions in Northeast China, accounting for 68.11% in areal extent (Figure 6b, d). The AGB of forestland is mainly stored in the barks, trunks, leaves, and branches of forest (trees and shrubs). The carbon sink functions of different forest types vary substantially. The increase of carbon of the same forest type also differs with the forest age (Hu et al. 2015a; Hu et al. 2015b). Due to the over-simplification of the carbon cycle in the InVEST model, the carbon fixation rate over a short period is assumed to be constant. The forestland carbon is thus often under- or over-estimated. Grassland and cultivated land accounted for 25.10% and 3.65% of the regional total carbon stock, respectively. Grassland and short-cycle (e.g. annual) crop carbon pools are relatively scarce because these pools renew too quickly or unsteadily (Guo et al. 2008). Most of the wetlands in Northeast China are bogs (acidic and dominated by shrubs) and/or fens (neutral to slightly alkaline and dominated by sedges (e.g. Carex tato) or other grasses) with an areal extent of 2.31% and a high carbon content (Mu et al. 2013b). Other LULC types have much less areal extents and lower carbon.

Figure 6. Forest age and land use/land cover (LULC) of permafrost region in Northeast China in the 2010s.

(1) A) Coniferous forest; B) Broad-leaved forest; C) Coniferous and broadleaf mixed forest; D) Shrubs; E) Grassland; F) Wetlands; G) Cultivated land; H) Water bodies; I) Construction land, and; J) Unused land.

(2) SLLP stands for the southern limit of latitudinal permafrost.

The carbon cycle of the forest ecosystem largely depends on the forest age structure (Yu et al. 2014). Forest age is an important parameter to evaluate the carbon sink potential of forest ecosystems at the regional scale (Wang et al. 2011; Zhou et al. 2016). There were significant differences in the carbon sink function for different forest ages (Zhou et al. 2006). The introduction of forest age can effectively improve the spatial accuracy of model simulation for the forest carbon cycle (Zhou et al. 2016). At present, in the estimation of carbon storage based on the InVEST model in Northeast China, forest carbon density is assigned as a fixed value without considering the influence of forest age, which leads to great uncertainty in simulating the forest carbon fixation (Li et al. 2022a; Ren, Cao, and Wang 2023). In this paper, forested land is reclassified by introducing the vegetation map. The dynamic relationships between forest age and carbon density in coniferous, broad-leaved, and coniferous-broadleaf mixed forests were established based on field forest investigations in Northeast China. In Figures 6a, c, the average forest age in permafrost regions is mainly 70–130 years. Forests have strong carbon fixation capacity and thus great carbon sink potential (Hu et al. 2015b). However, with increasing forest age, there are different understandings of the change in forest carbon sink capacity. Some studies have shown that the productivity of young forests is greater than that of old forests (Ryan, Binkley, and Fownes 1997), and some studies believe that old forests have a larger carbon sink function (Zhou et al. 2006). Therefore, there may be some uncertainties in the linear relationship between forest age and carbon density established in this paper. The next step is to establish a more accurate relationship between forest age and carbon density by conducting extensive surveys and combining forest inventory data. Taking Heilongjiang Province as an example, the forest carbon stocks (AGB and BGB) estimated in this paper were compared with the forest inventory (AGB and BGB). In this study, the forest carbon stocks in the late 1980s (1986), 2000, and 2010 were estimated at 550.73, 706.69, and 825.66 Tg, respectively. During 1984–1988, 1999–2003, and 2009–2013, the carbon stocks of forest inventory were 696.87, 702.41, and 833.99 Tg, respectively (Zhang et al. 2018). The results of this study underestimated forest carbon stocks in the late 1980s and approached that of forest inventory in 2000 and 2010. The carbon density of other LULC types was assigned based on reviewing field survey results in the existing literature for Northeast China. The carbon density parameters are downscaled locally (landscape to small basins) to regionally (different zones of permafrost) in Northeast China, and the accuracy of carbon estimation is thus effectively improved.

4.2. Effects of climate change and human activities on carbon stock in permafrost regions

Climate change affects the carbon cycle of terrestrial ecosystems by changing temperature, precipitation, and others (Ni et al. 2019). In the permafrost regions of Northeast China, the multi-year average of annual precipitation from 1980 to 2020 was 435.6 mm, and the mean annual air temperature was −2.0°C, all with increasing trends (Figure 7). There are significant differences in the sensitivity of various ecosystems to climate change (Oechel et al. 2000). For example, the carbon density of forests increases with forest age. However, climate warming may modify or change the dominant position of forest species and affect the carbon fixation capacity of forests (Dai et al. 2013). Climate change, especially in precipitation, is the main influencing factor of the grassland carbon pool (Xin et al. 2020). The impacts of climate change on the grassland carbon pool are far greater and more extensive than those of grazing (Xin et al. 2020). The contribution proportions of climate change to AGB and BGB of grassland are 35.4% and 18.7%, respectively (Xin et al. 2020). The total carbon stock in the wetlands in the Da and Xiao Xing’anling Mountains is about 1.58 Pg C (Xue et al. 2021). With the warming of the climate and the degradation of permafrost, the areal extents of boreal wetlands in the Da and Xiao Xing’anling mountains are shrinking. Nearly one-quarter of boreal wetlands would be lost, and over one-fifth of existing carbon pool would be at risk of destabilization by the 2070s under high emission scenarios (Xue et al. 2021).

Snow cover is the most active and sensitive environmental factor at middle and high latitudes. Snow cover has significant influences on hydroclimatic and ground hydrothermal regimes. Snow cover affects the accumulation, distribution, and degradation of organic matter (carbon) in soils (Stieglitz et al. 2003; Gouttevin et al. 2012; Jan and Painter 2020). Northeast China has regions with stable snow cover because of heavy snowfall, large snow depth, and long duration of snow accumulation. The average annual snow depth from 1980 to 2020 was 5.0 cm (Figure 7). Snow cover significantly affects the hydrothermal conditions and thaw depth of the Xing’an permafrost, potentially exposing more carbon-laden soil to microbial decomposition (Zhang 2005; Jan and Painter 2020).

Figure 7. Mean annual snow depth, annual precipitation and mean annual air temperature changes in permafrost region in Northeast China from 1980 to 2020.

Notes: The data were collected from 17 meteorological stations in the permafrost regions in the northern part of Northeast China

Forest fire has important influences on hydrothermal state, carbon and nitrogen cycles, and ecosystem successions in permafrost regions (Holloway et al. 2020; Li et al. 2021b). Forest fire results in increasing active layer thickness, rise of ground temperature, and decrease in soil moisture content in permafrost regions. Moreover, thermokarst development and ground-surface subsidence may occur in ice-rich permafrost regions under a warming climate and increasing fire occurrences and human activities (Li et al. 2019; Holloway et al. 2020). Forest fire induces the degradation of permafrost and changes in soil carbon and nitrogen budgets, and a large amount of SOC and soil nitrogen are released (Li et al. 2020; Jin and Ma 2021). For example, from 2001 to 2010, the total carbon emission from forest fires in the Da Xing’anling Mountains was 5.36 Tg C (Hu, Wei, and Sun 2012).

The Xing’an permafrost in loose sediments is generally warm (>–1°C) and ice-rich. Permafrost is thermally unstable and very sensitive to human activities (Jin et al. 2007; Li et al. 2021a). For example, the China–Russia crude oil pipelines run through primeval boreal forests and wetlands in Northeast China. The pipelines’ construction and operation exposed permafrost and caused water ponding in pipe trenches or the right-of-way, thus causing the melting of ground ice (Jin 2010). Engineering activities have contributed to the degradation of boreal forests and wetlands in the Xing’an permafrost region in Northeast China, contributing to carbon emissions (Li et al. 2021a).

In conclusion, the Xing’an permafrost region is located on the southern margin of the boreal forest belt and East Asian latitudinal permafrost regions. Forests, wetlands, and permafrost are interdependent and important ecosystem carbon pools in China. Due to the lack of long-term monitoring and detailed research, the carbon storage of terrestrial ecosystems in the Xing’an permafrost region is unclear. Based on multi-period LULC, vegetation map, forest age map, and literature data, the carbon storage of the terrestrial ecosystem in the Xing’an permafrost region was preliminarily estimated in this study. However, permafrost is also an important SOC pool for global terrestrial ecosystems (Zimov, Schuur, and Chapin 2006). It is estimated that the Northern Hemisphere permafrost SOC accounts for about 50% of the global SOC pool (Hugelius et al. 2014). Permafrost degradation results in the rapid decomposition of a large amount of long-sequestration SOC (Ding et al. 2017; Turetsky et al. 2019). Permafrost carbon release will not only profoundly change the carbon source and sink function of terrestrial ecosystems, but also significantly increase the concentration of major greenhouse gases, such as CH₄ and CO₂ in the atmosphere and exacerbate climate warming (Harden et al. 2012; Song et al. 2021). There is a close relationship between terrestrial ecosystems and spatial and temporal distributive patterns of near-surface permafrost. The hydrothermal processes of soils in the active layer affect the vegetation patterns, community composition and successions, and carbon exchange (Fisher et al. 2016). Therefore, future studies will focus on the interaction and feedback mechanisms between terrestrial surface ecosystem carbon storage and permafrost SOC, especially the SOC sequestered at greater depth (>0.3–3.0 to about 20.0 m), which are prone to permafrost thawing under a warming climate and increasing human activities.

5. Conclusions

In this study, the carbon estimation module of the InVEST model was improved by introducing the parameters of forest age and vegetation type. The total carbon stock in the regions of Xing’an permafrost in Northeast China was estimated for four periods in the late 1980s, 2000, 2010, and 2020. The carbon models were driven by the two scenarios of “changing LULC and constant carbon density” and “changing carbon density and constant LULC” to estimate the total carbon stock. Furthermore, the influences of the LULC transformation and vegetation carbon fixation on the carbon pool were evaluated. The results indicate a significant degradation of the permafrost from the 1980s to 2010s, with the areal reduction of permafrost by 1.086 × 10⁵ km². The total carbon stock in permafrost regions gradually declined from 4293.04 Tg C (woodland, grassland, and wetland account for 73.96%, 17.29%, and 6.84%, respectively) in the 1980s to 4049.56 Tg in the 2010s. Based on the areal extent of frozen ground regions of the 1980s, two scenarios were used in estimating the total carbon pools from the late 1980s to 2020. The reduction of carbon stock caused by LULC transformation (−33.53 Tg C) is less than the fixation of carbon due to vegetation growth (+1205.18 Tg C), resulting in a net gain of carbon stock at 1149.15 Tg C in Xing’an permafrost regions in Northeast China.

Current studies lack detailed vegetation maps spatiotemporally matched with LULC data. Therefore, the conversion between different forest types had to be ignored in the model calculation. Due to the lack of field investigation data on the carbon density of shrubs at different ages, the average carbon density of shrubs in Northeast China was used to assign input values of forest age in the model. This may result in an underestimation of the carbon stock of shrubs. In the future, we will combine the results from field surveys, remote sensing data, and model simulations to more accurately analyze the changes in carbon stock caused by conversion between forestlands. This will help to better estimate carbon exchanges between ecosystems and the active layer and near-surface permafrost from the regions of Xing’an permafrost. This study helps better evaluate the carbon pool and its stability, the carbon emission, and sink potential of boreal ecosystems, and the management of boreal forest and wetland ecosystems in Northeast China.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The simulation data of total carbon in the Xing’an permafrost regions in Northeast China used in this study are available through the Mendeley Data. https://dx.doi.org/10.17632/hrjmn2gbzd.1.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

This research is jointly supported by the National Natural Science Foundation of China (NSFC) (Grant Nos. 42101408, 42001052, and U20A2082), the Startup Research Funding of Northeast Forestry University for Chengdong Leadership (LJ2020–01), Fundamental Research Fund for the Central Universities (Grant No. 2572021DT08), the State Key Laboratory of Frozen Soils Engineering Open Fund (Grant No. SKLFSE202008), and the funding from the Autonomous Province of Bozen/Bolzano-Department for Innovation, Research and University in the frame of the Seal of Excellence Programme (project TEMPLINK, Grant No. D55F20002520003).