Next Article in Journal
Uncertainty Analysis of Forest Aboveground Carbon Stock Estimation Combining Sentinel-1 and Sentinel-2 Images
Previous Article in Journal
Moso Bamboo’s Survival Strategy Against Chilling Stress in Signaling Dynamics
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Forests
Volume 15
Issue 12
10.3390/f15122133
forests-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

An Improved Grid-Based Carbon Accounting Model for Forest Disturbances from Remote Sensing and TPO Survey Data

by
Weishu Gong
1,*,†,
Chengquan Huang
1,
Yanqiu Xing
2,*,
Jiaming Lu
3 and
Hong Yang
2
1
Department of Geographical Sciences, University of Maryland, College Park, MD 20742, USA
2
Centre for Forest Operations and Environment, College of Mechanical and Electrical Engineering, Northeast Forestry University, Harbin 150040, China
3
School of Biological Sciences, University of Utah, Salt Lake City, UT 84112, USA
*
Authors to whom correspondence should be addressed.
†
This work is part of the doctoral thesis of the first author Weishu Gong, Doctoral Programme, University of Maryland, College Park, United States.
Forests 2024, 15(12), 2133; https://doi.org/10.3390/f15122133
Submission received: 23 October 2024 / Revised: 24 November 2024 / Accepted: 29 November 2024 / Published: 2 December 2024
(This article belongs to the Section Forest Inventory, Modeling and Remote Sensing)
Browse Figures
Versions Notes

Abstract

:
Forest disturbance is one of the main drivers of forest carbon flux change. How to accurately estimate the carbon flux caused by forest disturbance is an important research problem. In a previous study, the authors proposed a Grid-based Carbon Accounting (GCA) model that used remote sensing data to estimate forest carbon fluxes in North Carolina from 1986 to 2010. However, the original model was unable to track legacy emissions from previously harvested wood products and was unable to consider forest growth conditions before and after forest disturbance. This paper made some improvements to the original GCA model to enable it to track fluxes between all major aboveground live carbon pools, including pre-disturbance growth and growth of undisturbed forests, which were not included in the initial model. Based on existing timber product output (TPO) survey data and annual TPO records inversed from remote sensing data, we also worked to clarify the distribution ratios of removed C between slash and different wood product pools. Specifically, the average slash ratio for North Carolina was calculated from the difference between the C removed and the C flowing into the wood product as calculated from TPO survey data. County- and year-specific ratios were then calculated using the annual TPO records obtained from remote sensing and TPO survey data, dividing the removed remaining C into pools P1, P10, and P100, which were then applied to each 30 m pixel based on the county and year to which the pixel belonged. After compensating for these missing legacy emissions and adjusting forest growth rates from Forest Inventory and Analysis (FIA) data, we estimated a net carbon sink of 218.1 Tg of the flux associated with live aboveground biomass and harvested wood products from North Carolina woodlands over the 25-year study period (1986–2010). This estimate is close to the greenhouse gas emission and sink data provided by the U.S. Department of Agriculture for North Carolina and is comparable to estimates reported by several other studies.

1. Introduction

Currently, there are great uncertainties in estimating the carbon budget of terrestrial ecosystems [1]. Forests are the largest carbon sources and sinks in terrestrial ecosystems, covering more than 4.1 billion hectares of forest, and the forest carbon budget contains more than 80% of aboveground and 40% of subsurface carbon [2]. Change in the forest carbon budget is largely influenced by forest disturbance and subsequent regeneration process control. Through disturbance, forest carbon is released into the atmosphere or transferred to other carbon pools (e.g., building materials, furniture, etc.), where the carbon is gradually released [3,4,5].
Forest disturbance (including disturbances caused by land use change, logging, fires, storm damage, insect outbreaks, etc.) is one of the main influences of forest carbon flux change [6,7,8,9]. How to accurately estimate the carbon flux caused by forest disturbance is a difficult research problem. In many studies, remote sensing data are used to map land cover and various surface features [10,11]. Multi-generational time-series observations provided by Landsat are particularly valuable for understanding carbon dynamics associated with land use change over the past few decades. In 2008, the Landsat archives were opened to the public, opening up new opportunities to map the history of land change using time-series Landsat observations. LandTrendr [12,13], Vegetation Change Tracker (VCT) [14], and Continuous Change Detection and Classification (CCDC) are among the various algorithms designed for land change studies using time-series Landsat observations.
Since the release of Landsat data, Sulla-Menashe et al. [15] have led efforts to characterize forest disturbance in the United States using Landsat observational time series. Francini et al. [16] combined Landsat and photonic (GEDI) data to monitor disturbances in Italian forests over four decades (1985–2019). Diao et al. [17] assessed the evolution of plantation forest stand age in southeastern China from 1987 to 2017 using VCT. These products were derived using the VCT algorithm and 30 years of surface reflectivity records. The VCT detects unusual events in the spectral time series per pixel caused by forest disturbance, including harvest/logging, fires, storm damage, and insect outbreaks [18,19]. With biannual Landsat observations, the accuracy of VCT in detecting stand clearance and partial disturbance was 92% and 60%, respectively [20].
Landsat time series also provide an opportunity to derive other important biophysics related to forest carbon dynamics [21,22]. Zhang et al. [23] mapped disturbance causality in boreal and arctic ecosystems of North America using time series of Landsat data. Tao et al. [24] used field measurements collected by the U.S. Forest Service’s Forest Inventory and Analysis (FIA) program as calibration data to estimate the percentage of base area removal (PBAR) as a measure of the disturbance intensity of disturbance events in North Carolina detected by the VCT algorithm. Zhang et al. [25] assessed spatio-temporal changes in forest cover and fragmentation under urban sprawl using Landsat long-term observation data (1987–2017) using the Vegetation Change Tracker (VCT) algorithm. In addition, VCT interference products can also obtain annual and multi-generational (1986–2015) records of forest-harvested wood timber product output (TPO) [26,27], as well as the inflow of harvested carbon into wood products [28]. Compared to TPO survey data [27], Landsat-based TPO records are much longer and provide yearly estimates. Gong et al. (2024) [29] used Landsat remote sensing and TPO data to estimate annual harvested wood products in seven states in the southeastern United States.
Recently, many studies have been conducted on carbon flux-related modelling. Law et al. [30] investigated the carbon impacts of drought, fire, and forest management, emphasizing both current and future effects on forests in the Pacific Northwest. Moomaw et al. [31] analyzed various scenarios of future timber supply and demand, using these to develop a carbon costing system that incorporates a discount rate for future carbon emissions and removals. Peng et al. [32] presented a new model that applies time discounting to estimate current and future carbon costs of global timber harvesting under different scenarios. Qing et al. [33] introduced a series of carbon flux network models based on complex network theory, quantifying the impact of carbon fluxes within China’s climate zones by incorporating a time-delay effect. The Grid-based Carbon Accounting (GCA) model developed by Gong et al. (2022) [34] provided a framework for estimating C fluxes over areas affected by forest disturbances at the 30 m resolution, which is needed to calculate the C fluxes arising from harvest or other disturbances over individual forest patches. Such patch-level C estimates are crucial for supporting various carbon management schemes, carbon trade, and other climate change mitigation initiatives at individual project and/or individual property owner levels [35,36]. To derive C estimates for all forest lands, however, the fluxes over forests not affected by contemporary disturbances should also be included.
Further, the annual TPO records derived in [29] as the main input data for the GCA model provided an opportunity to improve C flux estimates related to different wood product pools. Following the original Bookkeeping Carbon Accounting (BCA) model, the GCA model uses fixed fractions when dividing harvested wood products into different release pools. In reality, the release pool fractions of carbon sources in different grids could be different. Derived using satellite-based annual forest disturbance data and TPO survey data, the TPO records from [29] provide year- and county-specific estimates of fuelwood, pulpwood, and sawlog volumes. These records could be used to derive more realistic partitioning of the total biomass removed by disturbance events into the P1, P10, and P100 release pools (i.e., pools with C assumed to release in 1, 10, and 100 years, respectively).
Our current study had two major goals. The first was to further improve the GCA model such that it could produce flux estimates for pre-disturbance growth over disturbed areas, as well as the growth of undisturbed forests, and to use the TPO records [29] to derive more realistic partitioning of the total biomass removed by disturbance events into the P1, P10, and P100 release pools, leading to the calculation of a more realistic forest carbon budget. Other ecosystem carbon pools (belowground biomass, dead wood, litter, and soil) collectively make up the bulk of forest carbon stocks (about 5, 10, 25, and 40 percent, respectively) [37,38,39], but analyzing them is beyond the scope of this study, which focuses on the carbon budget of the live aboveground biomass and harvested wood products. Following a comparison of model outputs with available inventory-based flux estimates, the growth rates used by the model were also adjusted according to rates calculated using FIA field plot data. The second goal was to use the improved model to evaluate how forest carbon budget estimates might be affected by forest management practices.

2. Methods

2.1. Study Area

The study area was the entire state of North Carolina, located in the southeastern United States, which is considered one of the most frequently disturbed forests in North America. Following logging and agricultural expansion in the United States during the 18th and 19th centuries, the region experienced large-scale commercial wood harvesting in the 20th century [35], and by the mid-20th century, forest regrowth contributed to the high potential of the region to become a net carbon sink [35,40]. Today, this region contains two ecoregions, designated as Regions 21 and 35, which correspond to the two landscapes of humid subtropical forests and temperate mountain systems, respectively. There are nearly 100 counties spread between the eastern Atlantic coast and the western Great Smoky Mountains. North Carolina covers an area of 139,390 km2, of which approximately 60% (75,199.3 km2) is forested [41], much of which is categorized as woodland, and contains a very rich variety of forest types, such as oak–hickory, small-leaf–shortleaf pine, oak–pine, and oak–eucalyptus–cypress. Between 1985 and 2010, more than half of the state’s forests were disturbed at least once [27]. The types of disturbances are not limited to human timber harvest; hurricanes, insect invasions, forest fires, rain, snow, frost, and other types of disturbances from nature are also common. The total area of the state disturbed by forest stand clearing is relatively stable, but there are still some disturbance impacts that produce large inter-annual variability [24].

2.2. The Original Grid-Based Carbon Accounting (GCA) Model and Model Inputs

The original GCA model proposed by Gong et al. (2022) [34], derived from the original Bookkeeping Carbon Accounting (BCA) model [42], processes data in 30 m grids. It divides the study area into 30 m grids, runs the BCA model within each grid, and trains the model to determine its parameters, enabling 30 m grid-based carbon flux calculation. Since this study essentially reimplements the GCA model, most parameter values are equivalent to those used in the original GCA model. The remote sensing products used in the original GCA model primarily include a set of forest disturbance maps [27] and a pre-disturbance carbon density map obtained through spatial filtering of the carbon density map developed by Wilson et al. [43], both with a resolution of 30 m. For all other inputs, the default parameter values provided by the original BCA model were used [42].

2.3. GCA Model Improvement

The article [34] introduced the GCA model to estimate carbon flux from forest disturbance using remote sensing data. However, the forest disturbance in that paper only takes into account changes in land use and land cover; the deforested area is considered to be fully harvested. And, on the other hand, it does not take into account the usage of the deforested forest; consequently, it is difficult to accurately calculate carbon stocks. Considering all the abovementioned factors, the GCA model was improved in this paper to achieve more accurate carbon accounting.

2.3.1. Partitioning of Removed C by Disturbance Events

The aboveground carbon removed by disturbance events can be divided into four pools from which the C is released through different processes at different rates—slash (defined here as woody debris left on site), P1, P10, and P100 (Table 1). The original BCA model provided the partitioning ratios among the four pools for each ecozone, which were assumed to remain the same across counties and over time within each ecozone. In reality, however, these ratios could differ from county to county and vary from year to year. The TPO data provided an opportunity to derive county- and year-specific partitioning among the P1, P10, and P100 pools. Specifically, softwood and hard fuelwoods should go to the P1 pool because they likely will be burned in one year. Pulpwood might store C for up to a decade and should be in the P10 pool. Sawlogs could store C in furniture and buildings for over a century, and hence should be in the P100 pools. No direct estimates of slash (woody debris left on site) were available in the TPO survey data. To indirectly estimate residual carbon left on-site over a large area, we calculated slash as the difference between the total carbon removed from the trees during logging (Cvct) and the carbon stored in wood products (Ctpos), as shown in Equation (1). This approach assumes that all carbon not accounted for in wood products remains on-site as slash, which represents a simplification of reality but provides a practical method for large-scale estimation using available data.
Cslash = Cvct − Ctpos
where Cslash, Cvct, and Ctpos are the amounts stored in slash, the total C removed by disturbance events mapped by the VCT algorithm, and wood products, respectively. In any given year, Cvct for a county was calculated as the sum of the carbon removed at individual pixel locations (Cpj) for all pixels within that county (see Equation (2)).
Cvct = ∑ Cpj
For each TPO survey year, Ctpos was provided by the TPO survey data. Therefore, a slash ratio (Rslash) could be calculated using Equation (3) for each TPO survey year for every county.
Rslash = Cslash/Cvct
The county-specific Rslash values calculated this way, however, varied greatly among counties and TPO survey years. Assuming that the Rslash value was mostly determined by harvest methods, which were unlikely to differ that much among counties in the same state, the large among-county and among-TPO survey year differences were more likely caused by uncertainties in both the disturbance products and the pre-disturbance C data. To minimize the impact of such uncertainties, an average Rslash was calculated based on the county and TPO survey year-specific values. The resultant Rslash value, 0.4672, was then used to calculate the influx of C into the slash pool from the total C removed by each year’s disturbances for each county (CVslash) according to Equation (4).
CVslash = Cvct ∗ Rslash
After subtracting this slash pool from the total removed C, the P1, P10, and P100 pool fractions were calculated annually for each county by ratios of C in different wood products based on the annual TPO record derived in [34]. Finally, the C removed at each pixel location within a county was then partitioned according to the above derived county- and year-specific fraction values. Release of C stored in each product pool was calculated the same way as described in [34]. For a disturbance event that results in a disturbance area of S, the release from the carbon that ends up in the slash in any given year X is
Slashr in X = (Slash PoolX−1 + S ∗ CPrim ∗ FSlash) ∗ (1 − e−DRslash)
And the carbon pool after each year’s release is
Slash Pool in X = (Slash PoolX−1 + S ∗ CPrim ∗ FSlash) ∗ e−DRslash
where Slash PoolX−1 represents the carbon stored in the slash pool during the year prior to the disturbance (X − 1). For the wood products, based on different product types, the carbon pool is divided into a fast decay pool (10-year decay pool, or P10, e.g., paper products) and a slow decay pool (100-year decay pool, or P100, e.g., furniture and wood used in buildings), which release carbon annually at approximately 1/10th or 1/100th of their respective pool sizes. The DRSlash is the decay rate constant (Year−1), with values ranging from 0.01 to 0.5 depending on the ecozone. For this study, DRSlash values of 0.5 and 0.1 were applied to humid subtropical forests and temperate mountain systems, respectively, based on the reference range and ecozone-specific decomposition rates.

2.3.2. C Accumulation from Forest Growth

Forests transfer C from the atmosphere to the biomass pool through tree growth. Over disturbed areas, the GCA model developed in [34] used pre-defined growth rates to calculate the C accumulated by regrowing trees reestablished after each logging and/or fire disturbance event. The growth before the first disturbance event, however, was not tracked, nor was the growth over undisturbed forest areas. In this paper, both of these cases were included in calculating the carbon budget arising from forest growth.

2.4. Model Corroboration

The carbon budget of forest disturbance in North Carolina was recalculated according to the improved GCA model in the previous section. However, while the improved GCA model could produce C estimates at the 30 m spatial resolution, the validation of fine-resolution flux estimates remains a challenge. Flux estimates with spatial details close to the 30 m resolution are difficult to find in the published literature. Such estimates are often reported at global, national, or regional scales [44,45]. However, state-level details have been reported in some studies [44,45,46,47], which could be useful for evaluating the results derived in this study. In particular, the U.S. Forest Service published an inventory of greenhouse gas emissions and removals from forest land, woodlands, and urban trees [48]. This report provided annual flux estimates with state-level details from 1990 to 2018. Estimates were derived for both forest land that remained forest land and forest land conversions, and they included fluxes among aboveground biomass (including harvested wood products), belowground biomass, and other major carbon pools. Those related to aboveground biomass over forest land that remained forest land and forest land converted to urban land between 1990 and 2010 were used to evaluate the results derived through this study.

3. Results and Analysis

3.1. Improved Model’s Performance

Figure 1 shows a comparison of annual carbon flux estimates for the aboveground biomass pool, including harvested wood products, derived from the GCA model and the USFS inventory data over forest land that remained forest land in North Carolina. Overall, the aboveground biomass pool had large negative fluxes (sink) during the study period, and it was obvious that the GCA estimates were much larger than those derived from the USFS inventory data; the annual fluxes were twice as high as those obtained from the inventory data in the early 1990s and remained about 50% larger in 2010. As a result, the 21-year total sink estimate from the GCA model was 80% larger than the inventory-based estimate.
Compared to the relatively stable inventory data, the annual sink derived from the GCA model showed an obvious decreasing trend over the 21-year period. This is because the GCA model did not include legacy emissions from wood products harvested before 1986. As the model started to accumulate the wood product pool in 1986, the pool increased over time, and so did the model estimate of the legacy emissions from that pool, roughly following a logarithmic curve (Figure 2).
Assuming that all factors controlling the influx of C into the wood product pools and emissions from the pools between 1986 and 2010 would not change that much in future years, the fitted curve could be used to estimate the legacy emissions in any year after 1986 from wood products harvested between 1986 and that year. For example, by the 100th year after 1986, the total annual legacy emissions from wood products harvested in the 100 years after 1986 would be 6.8 Tg. Assuming that the same method could also be used to estimate the legacy emissions in any year between 1986 and 2010 from wood products harvested 100 years before that year, then it would be possible to adjust the flux value estimated by the GCA model for each year such that the value would include legacy emissions from wood products harvested during the 100 years before that year. Figure 1 shows that after this adjustment, the annual C sink values from the GCA model were still 2.5 Tg–4.5 Tg larger than the inventory data, indicating that other factors, e.g., inflated growth rates, could have also contributed to the larger sink estimates of the GCA model.
As shown in the previous study [34], the growth rate used by the model for young forests (fast growth) in ecozone 21 was close to the mean growth rate calculated from FIA plot data. However, that mean growth rate was likely overestimated, given that many of FIA plots appeared to have abnormally high (5–10 times higher than the mean) growth rates [34]. Such high values had a disproportionately large impact on the mean value, but they were errors because no trees could grow that fast. Because medians are typically less affected by extremely large or small values, the median growth rates from [34] were used to replace the original growth rates used by the GCA model for young forests (fast). After this adjustment, annual sink estimates from the GCA model were still mostly larger than the inventory data, but the differences were much smaller than those derived using the original growth rates (Figure 3). After compensating for legacy emissions from wood products harvested during the 100 years before each model year, the annual sink estimates became slightly smaller than the inventory data, with differences ranging from 1 Tg to 2 Tg (Figure 3). These differences were less than half of the 2.5 Tg–4.5 Tg differences derived above using the original growth rates in the GCA model, suggesting that the growth rates of forests across North Carolina were better represented by the FIA-based median values than the original growth rates used by the model. Therefore, the FIA-based median growth rates were used in the GCA with growth adjustment model in the remaining analysis in this paper.
Compared to the initial GCA model developed in [34], the improved model developed in this chapter not only included fluxes from growth of pre-disturbance forests and undisturbed forests, but also adjusted forest growth rates and the method for calculating the influx of harvest carbon into the slash and wood product pools. Therefore, both sources and sinks derived from the improved model were different from those derived using the initial GCA model developed in [34], even over the disturbed areas (Figure 4).

3.2. North Carolina’s Forest Carbon Dynamics

The improved GCA model was used to produce carbon budget estimates at the 30 m resolution for living forests in North Carolina. The results showed that without considering soil carbon, dead wood, litter, and belowground biomass, as well as the legacy emissions from wood products harvested before 1986 and the emissions from wood products harvested from 1986 to 2010 that will be released after 2010, North Carolina’s forest land was a net C sink of 218.1 Tg over the 25-year (1986–2010) study period. Major sources included logging (136.7 Tg), conversion of forest to urban land (6.9 Tg), and fire (2.1 Tg). About 55% of the forests in North Carolina remained undisturbed during the study period. The growth of these forests resulted in the largest sink (179.1 Tg). While logging resulted in the largest source, post-logging regrowth and the C removed by logging that was transferred to the harvested wood product pool and was not released by 2010 created a sink of 103.2 Tg, and the sink from post-fire growth was 0.7 Tg. Further, pre-disturbance forest growth over the disturbed areas resulted in a sink of 80.8 Tg. Figure 5 shows the contributions of each source and sink process to the total source and sink estimates over the 25-year study period.
On average, North Carolina’s forest ecosystem, including live aboveground biomass and harvested wood products, had an average net carbon budget of −8.7 Tg/year (sink) between 1986 and 2010. This reflects the combined effects of carbon absorption and release during this period. The average annual carbon absorption was 14.6 Tg/year, including 7.2 Tg from undisturbed forests, 4.1 Tg from post-disturbance regrowth and carbon stored in harvested wood products not released by 2010, and 3.2 Tg from pre-disturbance growth. At the same time, annual carbon releases included 5.5 Tg from logging, 0.08 Tg from fire, and 0.3 Tg from conversion to urban land. This estimate does not include carbon fluxes from belowground biomass, dead wood, litter, or soil pools, which were beyond the scope of this study. This sink estimate seems to be comparable to those reported in a few studies that provided flux estimates for North Carolina. For example, Zheng et al. [47] reported an average net C change of −9.7 Tg/year for forest areas in North Carolina that remained forested between 1992 and 2001. Without specifying a year range, Han et al. [46] estimated that the “current annual” C change in forest biomass was −8.5 Tg/year. While it is not clear how legacy emissions from wood products were analyzed in those studies, the net sink estimate from this study should be smaller if legacy emissions from all wood products harvested over 100 years are to be fully accounted for.

4. Discussion

While the improved GCA model provides a way to derive high-resolution source and sink estimates for forest land, those estimates could be affected by many factors. In [34], it was demonstrated that the use of remote sensing-based biomass and disturbance intensity data had a large impact on the estimation of fluxes driven by disturbance and post-disturbance recovery. The use of newly available TPO records developed in [29] and field-based growth rates also resulted in estimates that were different from the earlier model (Figure 4) and were closer to inventory data (Figure 1 and Figure 3). While forest management in general can improve carbon storage [49], different harvest and management practices have different C outcomes [50,51,52]. Other important factors to consider in deriving flux estimates include land cover dynamics and methods for expressing emissions from forest disturbances [53].
The improved GCA model provided an opportunity to examine the carbon outcomes of different logging intensities in North Carolina and how these outcomes are influenced by different emission expression methods. Four logging intensity scenarios were considered, including clear cut (disturbance intensity = 100%), observed disturbance intensity as represented by the data developed by Tao et al. [24], half of the observed disturbance intensity, and zero disturbance (i.e., no logging). To estimate the timing and magnitude of carbon emissions, we considered two emission expression methods: “prompt emission”, where carbon is released primarily in the year of logging due to slash burning or rapid decomposition, and “delayed emission”, where carbon is released gradually over a span of 10 to 100 years, depending on the decay rates of different pools [54]. While prompt emissions can be directly observed using remote sensing techniques, delayed emissions depend on decay processes, which vary across regions and environmental conditions. The combination of logging intensity and emission expression methods resulted in eight scenarios (Table 2).
As expected, clear cut resulted in the highest fluxes from logging, followed by logging at the mapped intensity, half of the mapped intensity, and zero intensity (no logging). Except for the zero intensity case, flux estimates derived using the prompt emission method were always higher than those derived using the delayed emission method (Figure 6) for each of the other three logging intensities. The annual fluxes derived using the delayed emission method for each of these three logging intensity levels exhibited an increasing trend over time. As discussed earlier, this was mainly because the GCA model started to accumulate the wood product pools in 1986 and did not include legacy emissions from wood products harvested before 1986. The increasing trends were mostly the result of a growing wood product pool, which then resulted in higher emissions from that pool as the number of years since 1986 increased. Fluxes derived using the prompt emission method showed larger temporal variability than those derived using the delayed emission method but did not exhibit obvious temporal trends.
When fluxes were calculated using the delayed emission method, the logging source estimate over the logging areas for the 25-year study period would be ~60 Tg more if those areas had clear cut instead of having the logging intensities mapped by Tao et al. [24]. In contrast, the estimate would decrease by ~60 Tg if the observed logging intensities were reduced by half. Furthermore, the estimate would decline to 0 if the logging intensity were reduced to 0 (Figure 7). Logging intensity had little impact on post-logging growth. The flux from logging at the observed intensity (scenario MD) and post-logging growth over the logging areas was a net source of 33.5 Tg over 25 years. That source would become 95.8 Tg if the logging practice were changed to clear cut. However, if the observed logging intensity were reduced by half, the sink from post-logging growth would exceed emissions from logging, resulting in a net sink of 28.6 Tg. The sink from the disturbed areas would reach 94.5 Tg should logging be eliminated during the study period.
The emission expression methods had no impact on the calculation of sink from post-logging growth. However, because the logging source estimates derived using the prompt emission method were substantially higher than those derived using the delayed emission method, logging source estimates derived using the prompt emission method for the three non-zero intensity cases exceeded estimates of the sink from post-logging growth, resulting in positive net flux estimates for all three logging intensity levels.
Due to the large sink estimates from pre-disturbance growth and the growth of undisturbed forests, the aboveground biomass and harvested wood products associated with North Carolina’s forest land had a net sink estimate over the 25-year study period regardless of the logging intensity levels or the emission expression method used. The estimated size of the sink, however, decreased substantially as logging intensity increased from no logging to clear cut, and the pace of decrease was faster when flux estimates were derived using the prompt emission method rather than the delayed emission method.
Our results show that the carbon budget estimates of prompt emissions calculated using the model without considering the subsequent flow of harvested standing timber are much higher than the carbon budget estimates of delayed emissions estimated by the GCA model combined with the TPO record data. Therefore, the carbon stock in the subsequent flow of harvested standing trees cannot be ignored. Our improved GCA model accounts for delayed emissions of carbon after logging by tracking carbon footprint data subsequent to harvesting standing timber, and this improved model generates estimates that are close to the GHG emissions and sink inventory data provided by the USFS for the state of North Carolina and are comparable to estimates reported in several other studies. Although many existing models have taken into account the subsequent flow of harvested wood, most of these models are based on process-based carbon cycle models and have obvious deficiencies in the statistics of disturbance data [55]. In addition, current estimates of carbon fluxes from live trees based on satellite data typically fail to adequately account for prompt emissions from subsequent carbon flows from live tree harvesting. Even if the subsequent flow of standing trees is involved in mapping the overall carbon flux, this part of the carbon stock is usually calculated separately [56]. Compared to currently used carbon flux estimation models, our improved GCA model provides high-resolution estimates of carbon sources and sinks for standing forests, remedies the lack of carbon footprint tracking components in the direct estimation of carbon fluxes from forest disturbances, and complements the carbon stock data that circulate after harvesting standing trees. The model provides a new technical tool for accurate estimation of carbon fluxes in forest ecosystems.

5. Conclusions

This study further improved the GCA model by introducing existing TPO survey data and annual TPO records derived by Gong et al. (2024) [29] into the model in an effort to improve the allocation of removed C between stands and different wood product pools, allowing the improved model to track carbon among all major aboveground live carbon pools, including both pre-disturbance growth and growth in undisturbed forests. While the GCA model was designed to use remote sensing-based products to estimate fluxes arising from contemporary disturbances and hence could not track legacy emissions from wood products harvested before the study period, the model did track emissions from wood products harvested during the study period, which provided a basis for estimating the amount of legacy emissions from wood products that were not accounted for by the GCA model. After compensating for these missed legacy emissions and adjusting forest growth rates based on FIA field plot data, the model produced estimates that were close to the greenhouse gas emission and sink inventory data provided by the USFS for North Carolina and comparable to estimates reported in several other studies.
Without considering soil carbon as well as the legacy emissions from wood products harvested before 1986 and the emissions from wood products harvested from 1986 to 2010 that will be released after 2010, North Carolina’s forest land was a net C sink of 218.1 Tg over the 25-year (1986–2010) study period. That sink could be smaller if all logging areas had clear cut and could be larger if the logging intensities could be reduced by half or down to zero. In general, the prompt emission method resulted in larger source estimates from disturbances than the delayed emission method and hence reduced the sink estimates at all disturbance intensity levels.
Future research will aim to further improve the model’s accuracy and universality. One aim is to include more forest disturbance types, such as storm, ice, and insect disturbances. Another direction would be to include other carbon pools not considered in this study, such as soil carbon, so that the model would still be applicable to other regions. In addition, the dead wood was not explicitly included in the model developed here. However, dead wood primarily acts as a carbon source, releasing CO₂ through microbial decomposition and insect activity, which plays a dominant role in its contribution to the carbon cycle. Although minor, under specific conditions, dead wood may also act as a carbon sink by absorbing CO2 or CH4 from the soil. We acknowledge that excluding dead wood is a limitation of this study, and future research should account for its gradual accumulation and decomposition to better capture its role in forest carbon dynamics.

Author Contributions

Conceptualization, W.G.; methodology, W.G.; writing—original draft preparation, W.G.; writing—review and editing, C.H., Y.X., J.L. and H.Y. All authors have read and agreed to the published version of the manuscript.

Funding

This study was made possible by the Laboratory of Environmental Model and Data Optima (EMDO, grant no. 2106313).

Data Availability Statement

Dataset available on request from the authors.

Acknowledgments

The authors sincerely thank the Department of Geographical Sciences of the University of Maryland for the additional support provided. Elizabeth Burrill and Justin Holgerson of the US Forest Service assisted with the access to the FIA plot data, which was made possible through special agreements with the FIA program. We thank Jack (Jianguo) Ma for his long-term support.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Houghton, R.A. Terrestrial Fluxes of Carbon in GCP Carbon Budgets. Glob. Change Biol. 2020, 26, 3006–3014. [Google Scholar] [CrossRef]
  2. Le Quéré, C.; Andrew, R.M.; Friedlingstein, P.; Sitch, S.; Hauck, J.; Pongratz, J.; Pickers, P.A.; Korsbakken, J.I.; Peters, G.P.; Canadell, J.G.; et al. Global Carbon Budget 2018. Earth Syst. Sci. Data 2018, 10, 2141–2194. [Google Scholar] [CrossRef]
  3. Mitchard, E.T.A. The Tropical Forest Carbon Cycle and Climate Change. Nature 2018, 559, 527–534. [Google Scholar] [CrossRef]
  4. Fan, L.; Wigneron, J.-P.; Ciais, P.; Chave, J.; Brandt, M.; Sitch, S.; Yue, C.; Bastos, A.; Li, X.; Qin, Y.; et al. Siberian Carbon Sink Reduced by Forest Disturbances. Nat. Geosci. 2023, 16, 56–62. [Google Scholar] [CrossRef]
  5. Pugh, T.A.M.; Lindeskog, M.; Smith, B.; Poulter, B.; Arneth, A.; Haverd, V.; Calle, L. Role of Forest Regrowth in Global Carbon Sink Dynamics. Proc. Natl. Acad. Sci. USA 2019, 116, 4382–4387. [Google Scholar] [CrossRef] [PubMed]
  6. Ma, Z.; Peng, C.; Zhu, Q.; Chen, H.; Yu, G.; Li, W.; Zhou, X.; Wang, W.; Zhang, W. Regional Drought-Induced Reduction in the Biomass Carbon Sink of Canada’s Boreal Forests. Proc. Natl. Acad. Sci. USA. 2012, 109, 2423–2427. [Google Scholar] [CrossRef] [PubMed]
  7. Kurz, W.A.; Dymond, C.C.; Stinson, G.; Rampley, G.J.; Neilson, E.T.; Carroll, A.L.; Ebata, T.; Safranyik, L. Mountain Pine Beetle and Forest Carbon Feedback to Climate Change. Nature 2008, 452, 987–990. [Google Scholar] [CrossRef] [PubMed]
  8. White, J.C.; Wulder, M.A.; Hermosilla, T.; Coops, N.C.; Hobart, G.W. A Nationwide Annual Characterization of 25 Years of Forest Disturbance and Recovery for Canada Using Landsat Time Series. Remote Sens. Environ. 2017, 194, 303–321. [Google Scholar] [CrossRef]
  9. Kasischke, E.S.; Turetsky, M.R. Recent Changes in the Fire Regime Across the North American Boreal Region—Spatial and Temporal Patterns of Burning Across Canada and Alaska. Geophys. Res. Lett. 2006, 33, 3006–3014. [Google Scholar] [CrossRef]
  10. Loveland, T.R.; Dwyer, J.L. Landsat: Building a Strong Future. Remote Sens. Environ. 2012, 122, 22–29. [Google Scholar] [CrossRef]
  11. Wulder, M.A.; Loveland, T.R.; Roy, D.P.; Crawford, C.J.; Masek, J.G.; Woodcock, C.E.; Allen, R.G.; Anderson, M.C.; Belward, A.S.; Cohen, W.B.; et al. Current Status of Landsat Program, Science, and Applications. Remote Sens. Environ. 2019, 225, 127–147. [Google Scholar] [CrossRef]
  12. Kennedy, R.E.; Yang, Z.; Cohen, W.B. Detecting Trends in Forest Disturbance and Recovery Using Yearly Landsat Time Series: 1. LandTrendr—Temporal Segmentation Algorithms. Remote Sens. Environ. 2010, 114, 2897–2910. [Google Scholar] [CrossRef]
  13. Kennedy, R.E.; Cohen, W.B.; Schroeder, T.A. Trajectory-Based Change Detection for Automated Characterization of Forest Disturbance Dynamics. Remote Sens. Environ. 2007, 110, 370–386. [Google Scholar] [CrossRef]
  14. Huang, C.; Goward, S.N.; Masek, J.G.; Thomas, N.; Zhu, Z.; Vogelmann, J.E. An Automated Approach for Reconstructing Recent Forest Disturbance History Using Dense Landsat Time Series Stacks. Remote Sens. Environ. 2010, 114, 183–198. [Google Scholar] [CrossRef]
  15. Sulla-Menashe, D.; Kennedy, R.E.; Yang, Z.; Braaten, J.; Krankina, O.N.; Friedl, M.A. Detecting Forest Disturbance in the Pacific Northwest from MODIS Time Series Using Temporal Segmentation. Remote Sens. Environ. 2014, 151, 114–123. [Google Scholar] [CrossRef]
  16. Francini, S.; D’Amico, G.; Vangi, E.; Borghi, C.; Chirici, G. Integrating GEDI and Landsat: Spaceborne Lidar and Four Decades of Optical Imagery for the Analysis of Forest Disturbances and Biomass Changes in Italy. Sensors 2022, 22, 2015. [Google Scholar] [CrossRef]
  17. Diao, J.; Feng, T.; Li, M.; Zhu, Z.; Liu, J.; Biging, G.; Zheng, G.; Shen, W.; Wang, H.; Wang, J.; et al. Use of Vegetation Change Tracker, Spatial Analysis, and Random Forest Regression to Assess the Evolution of Plantation Stand Age in Southeast China. Ann. For. Sci. 2020, 77, 27. [Google Scholar] [CrossRef]
  18. Ye, J.; Wang, N.; Sun, M.; Liu, Q.; Ding, N.; Li, M. A New Method for the Rapid Determination of Fire Disturbance Events Using GEE and the VCT Algorithm—A Case Study in Southwestern and Northeastern China. Remote Sens. 2023, 15, 413. [Google Scholar] [CrossRef]
  19. Ding, N.; Li, M. Mapping Forest Abrupt Disturbance Events in Southeastern China—Comparisons and Tradeoffs of Landsat Time Series Analysis Algorithms. Remote Sens. 2023, 15, 5408. [Google Scholar] [CrossRef]
  20. Thomas, N.E.; Huang, C.; Goward, S.N.; Powell, S.; Rishmawi, K.; Schleeweis, K.; Hinds, A. Validation of North American Forest Disturbance Dynamics Derived from Landsat Time Series Stacks. Remote Sens. Environ. 2011, 115, 19–32. [Google Scholar] [CrossRef]
  21. Jiang, F.; Sun, H.; Chen, E.; Wang, T.; Cao, Y.; Liu, Q. Above-Ground Biomass Estimation for Coniferous Forests in Northern China Using Regression Kriging and Landsat 9 Images. Remote Sens. 2022, 14, 5734. [Google Scholar] [CrossRef]
  22. Neto, M.M.; da Silva, J.B.; de Brito, H.C. Carbon Stock Estimation in a Brazilian Mangrove Using Optical Satellite Data. Environ. Monit. Assess. 2023, 196, 9. [Google Scholar] [CrossRef]
  23. Zhang, Y.; Woodcock, C.E.; Chen, S.; Wang, J.A.; Sulla-Menashe, D.; Zuo, Z.; Olofsson, P.; Wang, Y.; Friedl, M.A. Mapping Causal Agents of Disturbance in Boreal and Arctic Ecosystems of North America Using Time Series of Landsat Data. Remote Sens. Environ. 2022, 272, 112935. [Google Scholar] [CrossRef]
  24. Tao, X.; Huang, C.; Zhao, F.; Schleeweis, K.; Masek, J.; Liang, S. Mapping Forest Disturbance Intensity in North and South Carolina Using Annual Landsat Observations and Field Inventory Data. Remote Sens. Environ. 2019, 221, 351–362. [Google Scholar] [CrossRef]
  25. Zhang, Y.; Shen, W.; Li, M.; Lv, Y. Assessing Spatio-Temporal Changes in Forest Cover and Fragmentation under Urban Expansion in Nanjing, Eastern China, from Long-Term Landsat Observations (1987–2017). Appl. Geogr. 2020, 117, 102190. [Google Scholar] [CrossRef]
  26. Markowski-Lindsay, M.; Brandeis, C.; Butler, B.J. USDA Forest Service Timber Products Output Survey Item Nonresponse Analysis. For. Sci. 2023, 69, 321–333. [Google Scholar] [CrossRef]
  27. Huang, C.; Ling, P.-Y.; Zhu, Z. North Carolina’s Forest Disturbance and Timber Production Assessed Using Time Series Landsat Observations. Int. J. Digit. Earth 2015, 8, 947–969. [Google Scholar] [CrossRef]
  28. Ling, P.-Y.; Baiocchi, G.; Huang, C. Estimating Annual Influx of Carbon to Harvested Wood Products Linked to Forest Management Activities Using Remote Sensing. Clim. Change 2016, 134, 45–58. [Google Scholar] [CrossRef]
  29. Gong, W.; Huang, C.; Zhao, F.; Lu, J. Estimation of Annual Harvested Wood Products Based on Remote Sensing and TPO Survey Data. Geo-Spat. Inf. Sci. 2024, 1–13. [Google Scholar] [CrossRef]
  30. Law, B.E.; Waring, R.H. Carbon Implications of Current and Future Effects of Drought, Fire and Management on Pacific Northwest Forests. For. Ecol. Manag. 2015, 355, 4–14. [Google Scholar] [CrossRef]
  31. Moomaw, W.R.; Law, B.E. A Call to Reduce the Carbon Costs of Forest Harvest. Nature 2023, 620, 44–45. [Google Scholar] [CrossRef] [PubMed]
  32. Peng, L.; Searchinger, T.D.; Zionts, J.; Waite, R. The Carbon Costs of Global Wood Harvests. Nature 2023, 620, 110–115. [Google Scholar] [CrossRef] [PubMed]
  33. Qing, T.; Wang, F.; Du, R.; Dong, G.; Tian, L. The Influence Patterns of Carbon Flux in Different Climatic Zones in China —Based on the Complex Network Approach. Europhys. Lett. 2024, 146, 31002. [Google Scholar] [CrossRef]
  34. Gong, W.; Huang, C.; Houghton, R.A.; Nassikas, A.; Zhao, F.; Tao, X.; Lu, J.; Schleeweis, K. Carbon Fluxes from Contemporary Forest Disturbances in North Carolina Evaluated Using a Grid-Based Carbon Accounting Model and Fine Resolution Remote Sensing Products. Sci. Remote Sens. 2022, 5, 100042. [Google Scholar] [CrossRef]
  35. Birdsey, R.; Pregitzer, K.; Lucier, A. Forest Carbon Management in the United States. J. Environ. Qual. 2006, 35, 1461–1469. [Google Scholar] [CrossRef]
  36. Corbera, E.; Schroeder, H. Governing and Implementing REDD+. Environ. Sci. Policy 2011, 14, 89–99. [Google Scholar] [CrossRef]
  37. Kurz, W.; Shaw, C.; Boisvenue, C.; Stinson, G.; Metsaranta, J.; Leckie, D.; Dyk, A.; Smyth, C.; Neilson, E. Carbon in Canada’s Boreal Forest—A Synthesis1. Environ. Rev. 2013, 21, 4. [Google Scholar] [CrossRef]
  38. Turetsky, M.R.; Abbott, B.W.; Jones, M.C.; Anthony, K.W.; Olefeldt, D.; Schuur, E.A.G.; Grosse, G.; Kuhry, P.; Hugelius, G.; Koven, C.; et al. Carbon Release Through Abrupt Permafrost Thaw. Nat. Geosci. 2020, 13, 138–143. [Google Scholar] [CrossRef]
  39. Lu, X.; Gilliam, F.S.; Yue, X.; Wang, B.; Kuang, Y. Shifts in Above-Versus Below-Ground Carbon Gains to Terrestrial Ecosystems Carbon Sinks Under Excess Nitrogen Inputs. Glob. Biogeochem. Cycles 2023, 37, e2022GB007638. [Google Scholar] [CrossRef]
  40. Hill, A.P.; Field, C.B. Forest Fires and Climate-Induced Tree Range Shifts in the Western US. Nat. Commun. 2021, 12, 6583. [Google Scholar] [CrossRef]
  41. Brown, M.; Eastell, L.; Gl, D.; Park, H.; Road, A. Alternatives to Venting of Natural Gas—ANG Gas Capture to Reduce Emissions. Available online: http://members.igu.org/old/IGU%20Events/igrc/igrc-2014/papers/fo4-2_brown.pdf (accessed on 28 February 2024).
  42. Houghton, R.A.; Hackler, J.L.; Lawrence, K.T. The U.S. Carbon Budget: Contributions from Land-Use Change. Science 1999, 285, 574–578. [Google Scholar] [CrossRef] [PubMed]
  43. Wilson, B.T.; Woodall, C.W.; Griffith, D.M. Imputing Forest Carbon Stock Estimates from Inventory Plots to a Nationally Continuous Coverage. Carbon Balance Manag. 2013, 8, 1. [Google Scholar] [CrossRef] [PubMed]
  44. Houghton, R.A.; Nassikas, A.A. Global and Regional Fluxes of Carbon from Land Use and Land Cover Change 1850–2015. Glob. Biogeochem. Cycles 2017, 31, 456–472. [Google Scholar] [CrossRef]
  45. Woodall, C.W.; Coulston, J.W.; Domke, G.M.; Walters, B.F.; Wear, D.N.; Smith, J.E.; Andersen, H.-E.; Clough, B.J.; Cohen, W.B.; Griffith, D.M.; et al. The U.S. Forest Carbon Accounting Framework: Stocks and Stock Change, 1990-2016; U.S. Department of Agriculture: Washington, DC, USA; Forest Service: Washington, DC, USA; Northern Research Station: Newtown Square, PA, USA, 2015; p. NRS-GTR-154. [Google Scholar]
  46. Han, F.X.; Plodinec, M.J.; Su, Y.; Monts, D.L.; Li, Z. Terrestrial Carbon Pools in Southeast and South-Central United States. Clim. Chang. 2007, 84, 191–202. [Google Scholar] [CrossRef]
  47. Zheng, D.; Heath, L.S.; Ducey, M.J.; Smith, J.E. Carbon Changes in Conterminous US Forests Associated with Growth and Major Disturbances: 1992–2001. Environ. Res. Lett. 2011, 6, 014012. [Google Scholar] [CrossRef]
  48. Domke, G.M.; Walters, B.F.; Nowak, D.J.; Smith, J.; Ogle, S.M.; Coulston, J.W.; Wirth, T.C. Greenhouse Gas Emissions and Removals from Forest Land, Woodlands, and Urban Trees in the United States, 1990–2018; Resource Update: Lake Town, Japan; U.S. Department of Agriculture: Washington, DC, USA; Forest Service: Washington, DC, USA; Northern Research Station: Newtown Square, PA, USA, 2020; Volume 227, pp. 1–5. [Google Scholar] [CrossRef]
  49. Canadell, J.G.; Raupach, M.R. Managing Forests for Climate Change Mitigation. Science 2008, 320, 1456–1457. [Google Scholar] [CrossRef]
  50. Naudts, K.; Chen, Y.; McGrath, M.J.; Ryder, J.; Valade, A.; Otto, J.; Luyssaert, S. Europe’s Forest Management Did Not Mitigate Climate Warming. Science 2016, 351, 597–600. [Google Scholar] [CrossRef]
  51. Nunery, J.S.; Keeton, W.S. Forest Carbon Storage in the Northeastern United States: Net Effects of Harvesting Frequency, Post-Harvest Retention, and Wood Products. For. Ecol. Manag. 2010, 259, 1363–1375. [Google Scholar] [CrossRef]
  52. Van Deusen, P. Carbon Sequestration Potential of Forest Land: Management for Products and Bioenergy Versus Preservation. Biomass Bioenergy 2010, 34, 1687–1694. [Google Scholar] [CrossRef]
  53. Ramankutty, N.; Gibbs, H.K.; Achard, F.; Defries, R.; Foley, J.A.; Houghton, R.A. Challenges to Estimating Carbon Emissions from Tropical Deforestation. Glob. Change Biol. 2007, 13, 51–66. [Google Scholar] [CrossRef]
  54. Fearnside, P.M. Greenhouse Gases from Deforestation in Brazilian Amazonia: Net Committed Emissions. Clim. Chang. 1997, 35, 321–360. [Google Scholar] [CrossRef]
  55. Zhou, Y.; Williams, C.A.; Hasler, N.; Gu, H.; Kennedy, R. Beyond Biomass to Carbon Fluxes: Application and Evaluation of a Comprehensive Forest Carbon Monitoring System. Environ. Res. Lett. 2021, 16, 055026. [Google Scholar] [CrossRef]
  56. Harris, N.L.; Gibbs, D.A.; Baccini, A.; Birdsey, R.A.; de Bruin, S.; Farina, M.; Fatoyinbo, L.; Hansen, M.C.; Herold, M.; Houghton, R.A.; et al. Global Maps of Twenty-First Century Forest Carbon Fluxes. Nat. Clim. Chang. 2021, 11, 234–240. [Google Scholar] [CrossRef]
Forests 15 02133 g001
Figure 1. Comparison of annual C changes in the aboveground (including harvested wood products) pool derived using the GCA model and from the USFS inventory data over forest land that remained forest land. The slope of the increasing trend in the annual flux estimates of the GCA model was reduced after adjusting the estimates to account for legacy emissions from wood products harvested during 100 years before each model year.
Figure 1. Comparison of annual C changes in the aboveground (including harvested wood products) pool derived using the GCA model and from the USFS inventory data over forest land that remained forest land. The slope of the increasing trend in the annual flux estimates of the GCA model was reduced after adjusting the estimates to account for legacy emissions from wood products harvested during 100 years before each model year.
Forests 15 02133 g001
Forests 15 02133 g002
Figure 2. Because the GCA model did not include legacy emissions from wood products harvested before 1986 and started to accumulate the wood product pool in 1986, model estimates of legacy emissions from wood products increased as the number of years since 1986 increased, roughly following a logarithmic curve.
Figure 2. Because the GCA model did not include legacy emissions from wood products harvested before 1986 and started to accumulate the wood product pool in 1986, model estimates of legacy emissions from wood products increased as the number of years since 1986 increased, roughly following a logarithmic curve.
Forests 15 02133 g002
Forests 15 02133 g003
Figure 3. Comparison of annual C changes in the aboveground (including harvested wood products) pool derived using the GCA model with growth rates adjusted based on FIA plot data and from the USFS inventory data over forest land that remained forest land. After adjusting the estimates to account for legacy emissions from wood products harvested 100 years before each model year, the annual sink estimated by the GCA model became slightly smaller than the inventory-based estimates.
Figure 3. Comparison of annual C changes in the aboveground (including harvested wood products) pool derived using the GCA model with growth rates adjusted based on FIA plot data and from the USFS inventory data over forest land that remained forest land. After adjusting the estimates to account for legacy emissions from wood products harvested 100 years before each model year, the annual sink estimated by the GCA model became slightly smaller than the inventory-based estimates.
Forests 15 02133 g003
Forests 15 02133 g004
Figure 4. Comparison of cumulative carbon stores between the initial GCA model [34], and the improved model developed in this paper. The improvements resulted in higher carbon source estimates (a) and lower carbon sink estimates (b) for forest land disturbed by logging. For other disturbance types, such as conversion to urban land (no sink) and forest land affected by fire, similar patterns were observed. (a) represents cumulative carbon released (source) in logging-disturbed forest land, and (b) represents cumulative carbon retained (sink) in logging-disturbed forest land.
Figure 4. Comparison of cumulative carbon stores between the initial GCA model [34], and the improved model developed in this paper. The improvements resulted in higher carbon source estimates (a) and lower carbon sink estimates (b) for forest land disturbed by logging. For other disturbance types, such as conversion to urban land (no sink) and forest land affected by fire, similar patterns were observed. (a) represents cumulative carbon released (source) in logging-disturbed forest land, and (b) represents cumulative carbon retained (sink) in logging-disturbed forest land.
Forests 15 02133 g004
Forests 15 02133 g005
Figure 5. Contributions of different source and sink processes to the total source and sink over North Carolina from 1986 and 2010. (a) The contribution of different source processes to total sources in North Carolina. (b) The contribution of different sink processes to total sinks in North Carolina.
Figure 5. Contributions of different source and sink processes to the total source and sink over North Carolina from 1986 and 2010. (a) The contribution of different source processes to total sources in North Carolina. (b) The contribution of different sink processes to total sinks in North Carolina.
Forests 15 02133 g005
Forests 15 02133 g006
Figure 6. Annual carbon emissions derived using the delayed (a) and prompt (b) emission methods at the four disturbance intensity levels shown in Table 2.
Figure 6. Annual carbon emissions derived using the delayed (a) and prompt (b) emission methods at the four disturbance intensity levels shown in Table 2.
Forests 15 02133 g006
Forests 15 02133 g007
Figure 7. Total fluxes over the 25-year study period derived using the delayed (a) and prompt (b) emission methods at the four disturbance intensity levels shown in Table 2.
Figure 7. Total fluxes over the 25-year study period derived using the delayed (a) and prompt (b) emission methods at the four disturbance intensity levels shown in Table 2.
Forests 15 02133 g007
Table 1. Ecozone-specific parameters used by the BCA model.
Table 1. Ecozone-specific parameters used by the BCA model.
NameUnitDescription
CPrimg/HaCarbon density in undisturbed, mature/primary forest
CMing/HaMinimum carbon density after disturbance
CSecg/HaCarbon density in recovered, secondary forest
FSlash Fraction of carbon ended up in slash pool
DRSlashYear−1Decay rate coefficient for slash pool
FP1 Fraction of carbon ended up in 1-year decay pool
FP10 Fraction of carbon ended up in 10-year decay pool
FP100 Fraction of carbon ended up in 100-year decay pool
TmsYearTime for forest to grow into secondary forest from stand-clearing disturbance
TspYearTime for forest to grow into mature forest from secondary forest
Table 2. Scenarios for assessing the impact of logging intensity and emission expression methods on forest C outcome estimates.
Table 2. Scenarios for assessing the impact of logging intensity and emission expression methods on forest C outcome estimates.
Disturbance IntensityDelayed ReleasePrompt Release
Clear Cut (100%)CDCP
Mapped Intensity [24]MDMP
Half of Mapped IntensityHDHP
Zero IntensityZDZP
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Gong, W.; Huang, C.; Xing, Y.; Lu, J.; Yang, H. An Improved Grid-Based Carbon Accounting Model for Forest Disturbances from Remote Sensing and TPO Survey Data. Forests 2024, 15, 2133. https://doi.org/10.3390/f15122133

AMA Style

Gong W, Huang C, Xing Y, Lu J, Yang H. An Improved Grid-Based Carbon Accounting Model for Forest Disturbances from Remote Sensing and TPO Survey Data. Forests. 2024; 15(12):2133. https://doi.org/10.3390/f15122133

Chicago/Turabian Style

Gong, Weishu, Chengquan Huang, Yanqiu Xing, Jiaming Lu, and Hong Yang. 2024. "An Improved Grid-Based Carbon Accounting Model for Forest Disturbances from Remote Sensing and TPO Survey Data" Forests 15, no. 12: 2133. https://doi.org/10.3390/f15122133

APA Style

Gong, W., Huang, C., Xing, Y., Lu, J., & Yang, H. (2024). An Improved Grid-Based Carbon Accounting Model for Forest Disturbances from Remote Sensing and TPO Survey Data. Forests, 15(12), 2133. https://doi.org/10.3390/f15122133

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop