Interactive comment on “ Estimating the carbon dynamics of South Korean forests from 1954 to 2012 ”

Forests play an important role in the global carbon (C) cycle, and the South Korean forests also contribute to this global C cycle. While the South Korean forest ecosystem was almost completely destroyed by exploitation and the Korean War, it has successfully recovered because of national-scale reforestation programs since 1973. There have been several studies on the estimation of C stocks and balances over the past decades in the South Korean forests. However, a retrospective long-term study that includes biomass and dead organic matter C and validates dead organic matter C is still lacking. Accordingly, we estimated the C stocks and their changes of both biomass and dead organic matter C during the 1954–2012 period using a process-based model, the Korean Forest Soil Carbon model, and the 5th South Korean national forest inventory (NFI) report. Validation processes were also conducted based on the 5th NFI and statistical data. Simulation results showed that the biomass C stocks increased from 36.4 to 440.4 Tg C at a rate of 7.0 Tg C yr −1 during the period 1954–2012. The dead organic matter C stocks increased from 386.0 to 463.1 Tg C at a rate of 1.3 Tg C yr −1 during the same period. The estimates of biomass and dead organic matter C stocks agreed well with observed C stock data. The annual net biome production (NBP) during the period 1954–2012 was 141.3 g C m −2 yr −1 , which increased from −8.8 g C m −2 yr −1 in 1955 to 436.6 g C m −2 yr −1 in 2012. Because of the small forested area, the South Korean forests had a comparatively lower contribution to the annual C sequestration by global forests. In contrast, because of the extensive reforestation programs, the NBP of South Korean forests was much higher than those of other countries. Our results could provide the forest C dynamics in South Korean forests before and after the onset of reforestation programs.


Introduction
Forests contain about 1146 Pg carbon (C), which consists of 359 Pg C in the vegetation and 787 Pg C in soils, and the global C sink of the forests was estimated to be 2.4±0.4Pg C yr −1 (Dixon et al., 1994;Pan et al., 2011) the function of forests as a C sink and their role in the global C dynamics has been recognized (IPCC, 2003;UNFCCC, 1997).Consequently, studies on the C budget of forest biomass and dead organic matter (DOM) have been conducted to understand temporal forest C stocks and balances (Bellassen et al., 2011;Kurz and Apps, 1999;Luyssaert et al., 2010;Pan et al., 2011;Piao et al., 2012;Stinson et al., 2011;Wang et al., 2007).
Furthermore, to display the net C changes in forest ecosystems, the net biome production (NBP), defined as net ecosystem production (NEP) minus disturbance loss or leaching, was also estimated (Luyssaert et al., 2010;Stinson et al., 2011).
To estimate the C stock and balance in forests, inventory-based estimation has been generally used because it could estimate C stock and net C balance directly.However, it has some limitations such as: not giving annual C budget, not necessarily taking into DOM C, and limitation on extrapolation due to high spatial variability (Piao et al., 2012;Wang et al., 2007).Recently, process-based modeling has been used for long-term simulation to provide the annual C budget of forests and to estimate C budget beyond the investigated area (Bellassen et al., 2011;Stinson et al., 2011;Wang et al., 2007).
The C dynamics of South Korean forests has varied largely.South Korean forests experienced severe deforestation over the 35 years of Japanese colonization  and the subsequent Korean War (1950War ( -1953) ) (Kang, 1998;Tak and Wood, 2007).Since 1973, following these periods of serious deforestation, the South Korean government implemented national plantation programs for the recovery of forests.
After about 30 years of effort, South Korean forests have successfully recovered and the stocking volume had increased from 8.2 in 1954 to 125.6 m 3 ha −1 in 2010 (Korea Forest Service, 2000, 2011).
Studies on the C stocks and balances in South Korean forests over the past decades have been conducted for many years.Based on the national forest inventory (NFI) and statistical data, the biomass C stocks of South Korean forests over the past decades were estimated (Choi and Chang, 2004;Fang et al., 2014;Li et al., 2010) rean forests were estimated (Piao et al., 2012;Yoo et al., 2013).However, there are some limitations such as: (1) relatively short simulation period, less than 20 years, (2) estimating C fluxes, not stocks, and (3) insufficient validation of DOM C. The primary objective of this study was to estimate the C stocks and annual C balance of South Korean forests, including biomass and DOM during the post-war period , using the Korean Forest Soil Carbon model (KFSC;Yi et al., 2013) and the 5th South Korean NFI as input data.To estimate the effect of reforestation programs, we provided the C dynamics before and after the onset of those programs.The estimated biomass and DOM C stocks were validated by comparing with the observed data in 5th NFI and Statistical Yearbook of Forestry.Furthermore, we compared the estimated annual C sink and NBP of South Korean forests with those of major countries and global forests.

5th NFI data
We used the 5th South Korean NFI data to prepare input data for the KFSC model and to validate the estimated DOM C stocks.The latest 5th NFI applied systematic cluster sampling for surveys and obtained data from about 4000 plots during 2006-2010(Korea Forest Research Institute, 2011).It provides information about forest type, species composition, diameter at breast height (DBH), age-class, stand density, topographical factors, observed C stocks of pools, and other data of each sampling plot.As we excluded denuded and bamboo forests in the simulation, 3890 plots (5 870 300 ha) were selected from the entire South Korean forests (6 368 843 ha).Each sampling plot represented 100, 400, or 1600 ha of forest grid cell, which is a simulation unit.Introduction

Conclusions References
Tables Figures

Back Close
Full 2.2 KFSC model

Model description
The KFSC model is an empirical and dynamic soil C model that consists of the five biomass compartments (stem, branch, foliage, coarse root, and fine root), five primary dead organic matter (DOM) compartments (aboveground woody debris from stem (AWDS), aboveground woody debris from branch (AWDB), aboveground litter (ALT), belowground woody debris (BWD), and belowground litter (BLT)), and three secondary DOM compartments (aboveground humus (AHUM), belowground humus (BHUM), and soil organic C (SOC)) classified according to the degree of decomposition and kinetics (Fig. 1).This model simulates forest C processes as follows: atmospheric C is converted to biomass, biomass becomes input to soils as litter and woody debris, litter and woody debris are decayed to humus (HUM), HUM is decayed to SOC, and SOC is decayed to CO 2 (Yi et al., 2013).Harvest is considered to be the only disturbance in the model.The performances of the KFSC model are described and validated in Park et al. (2013) and Yi et al. (2013).We parameterized the model for three needleleaf species (Pinus densiflora, P. rigida, and Larix kaempferi) and three broadleaf species (Quercus variabilis, Q. mongolica, and Q. accutissima).
To simulate the biomass C stocks, we followed the processes as follows: estimation of the growth of stemwood volume, conversion of stemwood volume to C stocks, and estimation of C stocks of other biomass compartments (branch, foliage, coarse root, and fine root).First, based on a yield table (Korea Forest Service, 2009), the growth functions of stemwood volume for each species and site index were parameterized using the Gompertz function (Table A1).An observed stemwood volume was assumed to follow the nearest stemwood growth function and site index.To calibrate the difference between the estimated stemwood volume from a selected growth function and the observed stemwood volume in the 5th NFI data for each stand, we multiplied the growth modifier used in Yi et al. (2013) 2013) used the ratio of fine roots to foliage described by Vanninen et al. (1996) as shown in Eq. ( 2): Fine roots : foliage ratio 0.0016 For the needleleaf species, this ratio was multiplied with the C stocks of the foliage to estimate the C stocks of fine roots.For the broadleaf species, we used the ratio of fine roots to coarse roots as 11 : 89 (Millikin and Bledsoe, 1999).By multiplying this ratio with the estimated C stocks of the coarse roots, we estimated C stocks of the fine roots for broadleaf species.

Input data and parameters
The required input data consisted of representative species, site index, growth modifier, forest age, and mean air temperature of each grid cell.The representative species was Figures

Back Close
Full determined as the tree species occupying the largest basal area (m 2 ) in each sampling plot.We used the forest age of each stand from the 5th NFI tree-ring data.For the plots without tree-ring data, the forest age was assumed to be 5, 15, 25, 35, 45, and 55 for age class I, II, III, IV, V, and VI, respectively, as reported in the 5th NFI data.
The mean air temperature of each grid cell was the average of the observed mean annual temperature from 1971 to 2000 from 75 weather stations over South Korea.
It was interpolated with a 0.01 • (∼ 1 km) grid size by the Kriging method, taking into consideration of the temperature lapse rate by elevation (Lee et al., 2007;Choi et al., 2011).The turnover rates of biomass C pools and decay rates of DOM C pools are required to simulate forest C processes (Table 1).The decay rate of ALT and turnover rates of branch and foliage were estimated by 2 years of field work data from 54 plots throughout South Korea (Lee et al., unpublished).The others were cited from other studies (Kim, 2002;Kurz et al., 1992;Liski et al., 2005;Noh, 2011;Park et al., 2006Park et al., , 2010;;Yoon et al., 2011).A detailed description of the modeling processes is given in Yi et al. (2013).

Model initialization and simulation
The method of reconstructing the forest age distribution is important for accurate simulation (Bellassen et al., 2011;Wang et al., 2007).Due to the lack of age information over the past decades, the forest C dynamics of each grid cell in South Korea during 1954-2012 were simulated by two scenarios, the spin-up scenario and the forest recovery scenario, reconstructing the forest age of each grid cell based on the recent age information in the 5th NFI.The spin-up scenario was applied to some stable stands during the simulation period and the recovery scenario was applied to most stands, specifically those that experienced severe deforestation due to exploitation and war.
The forest grid cells simulated by the spin-up scenario were initialized by a spin-up process until the iteration of a quasi-steady-state, such that the difference between Introduction

Conclusions References
Tables Figures

Back Close
Full SOCs at the end of two successive clear-cut rotations was < 1 % (Fig. 2a).After that, the model simulated forest C processes by forest age in 2012.In forest recovery scenario, those grid cells that experienced severe deforestation were also initialized by the spin-up process, while all DOM C pools, except SOC, were assumed to be zero in 1954, taking into consideration of land degradation caused by severe deforestation (Fig. 2b).Then, regeneration was assumed to be suppressed until the recent regeneration.SOC continued to decay slowly until the vegetation recovered sufficiently.As the vegetation recovered, biomass and DOM input started increasing.The forest grid cells simulated by the spin-up scenario were determined by the following criteria: (1) the stands regenerated before 1954 (over 60 years old), or (2) a stand located at the highest elevation for each age and province (Gangwon, Chungbuk, Chungnam, Chonnam, and Gyungbuk) from which the highest volume was harvested during 1970-2010(Korea Forest Service, 1974, 1988, 2000, 2005, 2010, 2012).These criteria assumed that stands that were old and located at high altitudes were not disturbed by exploitation and war.The other stands were assumed to had been severely destroyed, and were thus simulated by the forest recovery scenario.

Calculation of forest C stocks and NBP
To calculate the C stocks of the biomass and DOM C pools in South Korean forests, we applied Eq. ( 3): where, k is the identification number of each sampling plot, mean C density is the simulated biomass or DOM C per hectare, and A is the size of the grid cell, including each sampling plot.
The NBP (g C m −2 yr −1 ) of South Korean forests was also calculated.The NBP equals the net primary production minus the heterotrophic respiration minus the disturbance Introduction

Conclusions References
Tables Figures

Back Close
Full (fire, harvest, pests, land-use change, and other disturbances).As harvest was assumed to be the only disturbance and land-use change was not considered in this study, the KFSC model could simulate the net change of C stocks in the forest biome.
To calculate the NBP of South Korean forests, the annual change of C stocks in South Korean forests was divided by the total simulation area (5 870 300 ha).

Model validation
We validated the estimated biomass and DOM C stocks by comparing the estimated data to observed data from the statistical data and the 5th NFI data.For biomass, the estimates of stocking volume on a national scale during 1954-2010 were compared to the Statistical Yearbooks of Forestry (Korea Forest Service, 2000, 2011) to indirectly validate the estimated biomass C. Since observed data for DOM C from the past does not exist, the DOM C stocks in the 5th NFI data were used to validate the estimates.The estimates of soil C (BHUM + SOC) were multiplied by 0.6 (Lee et al., 2009) for comparison with the observed data, which was sampled at 0-30 cm depths.The estimates of other DOM values, those other than AWDS, BWD, and BLT (data is unavailable in the 5th NFI), were added to those of soil C for validation.The performance of the model in predicting DOM C stocks was analyzed using the root mean square error (RMSE).

The C stocks and annual C balances of biomass
An increase in C stock of biomass in South Korean forests was observed with the data increasing from 36.4 Tg C in 1954 to 440.4 Tg C in 2012, respectively (Fig. 3).

BGD Introduction Conclusions References
Tables Figures

Back Close
Full The estimated stocking volumes simulated by the KFSC model were compared with observed stemwood volume data to indirectly validate the biomass C stocks.The time series of estimated stocking volume showed a similar trend to that of observed stocking volumes on a national scale (r 2 = 0.98; Fig. 4).According to Statistical Yearbook of Forestry (Korea Forest Service, 2000, 2011), the stocking volume in South Korean forests increased from 51.8 to 800.0 Mm 3 between 1954 and 2010.The simulation result showed that it increased from 78.4 to 798.0 Mm 3 during that period.This implies the successful reconstruction of age distribution on a national scale, while it was still uncertain on a stand scale.
Our finding was consistent with other studies showing a large increase in biomass C stocks after the onset of reforestation programs.However, the mean biomass C density, biomass C stock, and annual C balance in this study were estimated higher than other studies (Table 2).There are two possible reasons explaining these differences.
As shown in Fig. 4, the stocking volume simulated by the model in recent years was an overestimate compared to the observed stocking volume.This caused the higher estimates of recent biomass C density and stocks, and annual C sink.The other possible reason is a difference in the methods of biomass C stock estimation.We estimated the biomass C stock with growth functions and BCFs.In contrast, Li et al. (2010) and Choi and Chang (2004) estimated the biomass C stocks by multiplying stemwood volume with constant biomass expansion factors (BEF).Variable BEFs could overestimate biomass C of a young forest compared to constant BEFs (Guo et al., 2010).As South Korean forests are relatively young, the estimated biomass C stocks of South Korean forests with BCFs could be higher than the estimates with constant BEFs.As the ratio of each compartment in biomass varied with stand age, our estimates were considered more realistic.

The C stocks and annual C balances of DOM
An increase in the C stock of DOM in South Korean forests was also observed, with the data increasing from 386.0 Tg C in 1954 to 463.1 Tg C and in 2012, respectively (Fig. 3).Figures

Back Close
Full As the forest vegetation had been almost denuded, the decomposition of the DOM C exceeded the organic matter input during that period.After the onset of reforestation programs and the recovery of litter input, the C stocks of DOM changed from a C source to a C sink sequestering C at a rate of 2.3 Tg C yr −1 .Averaged over the entire 1954-2012 period, the annual C balance of DOM was 1.3 Tg C yr −1 .The estimates and NFI inventories for DOM C stocks were in partial agreement (Fig. 5).The RMSE of the estimates were 26.9 and 49.2 Mg C ha −1 for needleleaf species and broadleaf species, respectively, on a regional scale.The underestimation of DOM C stocks could be partially explained by the mean air temperature used as input data.As recent air temperature has been higher than that of past centuries (Aizebeokhai, 2009), the decay rates of DOM C stocks might be overestimated for initialization process.Accordingly, the initial DOM C stocks were probably underestimated and uncertainties in estimating DOM C stocks occurred (Peltoniemi et al., 2006;Wutzler and Reichstein, 2007).Soil type also could affect the decay rate of HUM, ultimately the DOM C dynamics.Because of difference in dominant soil type, the DOM C stocks of Jeju province might be especially highly underestimated (Fig. 5).Separated from the mainland provinces, Jeju province is a volcanic island and the representative soil type of Jeju is Andisol (Ahn and Chon, 2010).Because of the strong association with allophane, DOM in Andisol soil is more stable and the decay of DOM C pools is suppressed (Calabi-Floody et al., 2011;Theng and Yuan, 2008).The overestimated decay rate of DOM C pools could underestimate the DOM C stocks in Jeju province.Excluding Jeju province, the RMSE improved to 12.8 and 21.9 Mg C ha −1 in needleleaf and broadleaf species, respectively.
As Jeju province accounts for only 1.8 % of South Korean forests, the estimated C stocks might be reliable.Introduction

Conclusions References
Tables Figures

Back Close
Full  during 1954-1973 and 1974-2012, respectively.As the forest vegetation recovered, the contribution of biomass to NBP increased over the simulation period and became higher than the that of DOM over that period.The contribution of biomass to NBP increased from 22.9 to 76.1 % during 1954-2012, and DOM accounted for the remainder.Although the C stock of DOM was also a C sink, that of biomass became more important C sink recently in South Korean forests.We compared the annual C balance and NBP of South Korean forests with those of forests from other countries and overall global forests (Table 3).Global forests annually sequestered about 3.8 Pg C yr tries was also lower than that of South Korean forests.This large difference in NBP might be attributed to extensive reforestation program in a national scale.

Uncertainties
Although we estimated the C budget and balance of South Korean forests, including biomass and DOM C, there are still uncertainties in the estimation.A site index, which is important input data for determining the productivity of a forest, seemed to be responsible for the uncertainty.In the KFSC model, the site index is determined by the forest age and observed stemwood volume, based on the yield table.While, in other process-based models, physiological processes were coupled to simulate the growth of biomass and various input data (e.g.temperature, CO 2 concentration, solar radiation, precipitation) are required (Chen et al., 2000;Ito et al., 2005;Krinner et al., 2005;Sitch et al., 2003).Using yield tables for a regional scale (Kurz et al., 2009) or quantifying the site index based on environmental factors (Nothdurft et al., 2012;Wang and Klinka, 1996) will help constrain the uncertainties associated with estimating a site index.To enable more precise assessment of South Korean forest C cycles, some important influences on C balance, such as CO 2 fertilization (Bellassen et al., 2011;Luyssaert et al., 2010), N deposition (Luyssaert et al., 2010), leaching (Luyssaert et al., 2010;Piao et al., 2012), forest area changes (Liski et al., 2006;Nabuurs et al., 2003), and management and other disturbances (Jandl et al., 2007;Kurz et al., 2009;Liu et al., 2002;Luyssaert et al., 2010;Zhou et al., 2013) need to be taken into consideration to understand comprehensive forest C dynamics.Technol., 38, 484-488, 2004. Choi, S., Lee, W. K., Kwak, D. A., Lee, S., Son, Y., Lim, K. H., and Saborowski, J.: Predicting forest cover changes in future climate using hydrological and thermal indices in South Korea, Clim. Res., 49, 229-245, 2011.Introduction

Conclusions References
Tables Figures

Conclusions References
Tables Figures

Back Close
Full  Full  Full Discussion Paper | Discussion Paper | Discussion Paper | While this approach could determine the net C change, DOM C stocks were excluded due to the lack of observed DOM C stock data.Using model, the C balances of South Ko-Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | The DOM C dynamics are the same as described by Yi et al. (2013); the detailed description of the DOM C dynamics in the KFSC model is given in Yi et al. (2013).
Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Before the onset of reforestation programs, the C stock of DOM was C source releasing C at a rate of 0.7 Tg C yr −1 .Until around 1980, the DOM C stocks were a C source.
Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 3.3 The total C budget and the NBP of South Korean forests Increasing total C stocks were observed.The C stocks in South Korean forests increased from 422.4 Tg C in 1954 to 903.5 Tg C in 2012, respectively.As the C emission from DOM C stocks overwhelmed C sequestration by biomass C stocks in the first two decades (1954-1973), South Korean forests were a C source and released C at a rate of 0.5 Tg C yr −1 during the period.From 1974, South Korean forests changed from C source to C sink sequestering C at a rate of 12.6 Tg C yr −1 .Averaged over the entire 1954-2012 period, the annual C balance was 8.3 Tg C yr −1 .Compared to the national fossil fuel-based C emissions data during 1954-2008 (Boden et al., 2012), South Korean forests annually offset 13.4 % of the South Korean fossil fuel C emissions during that period.The time series of NBP increased over the simulation period and the change in NBP showed an upward trend (Fig. 6).The NBP during 1954-2012 was 141.3 g C m −2 yr −1 and that in 1955 and 2012 were estimated to be −8.8 and 436.6 g C m −2 yr −1 , respectively.The onset of reforestation programs influenced the mean NBP and the averaged NBP was −8.7 and 214.4 g C m −2 yr −1 ,

− 1 (
Pan et al., 2011) and South Korean forests accounted for less than 1 % of that (8.3 Tg C yr −1 ).However, the NBP of South Korean forests exceeded that of foreign forests, and the global average, significantly.The NBP of global forests was around 100 g C m −2 yr −1 during 1990-2007 and that of South Korean forests during 1990-2007 was 365.2 g C m −2 yr −1 .The NBP of other major coun-Discussion Paper | Discussion Paper | Discussion Paper | using a model, we estimated the C dynamics of South Korean forests between 1954 and 2012.During this period, the C stocks of South Korean forests increased from 422.4 to 903.5 Tg C. South Korean forests changed from a C source Discussion Paper | Discussion Paper | Discussion Paper | to a C sink because of the extensive reforestation.The average annual C sink rate during this period was 8.3 Tg C yr −1 and the NBP was 141.3 g C m −2 yr −1 .From 1954-2008, 13.4 % of the fossil fuel C emission from Korea was offset by C accumulation in forest ecosystems.Although South Korean forests sequestered less C than other countries, the NBP was much higher.The high NBP is a result of extensive reforestation programs; thus, global-scale reforestation would contribute to C sequestration for the mitigation of global climate change.Discussion Paper | Discussion Paper | Discussion Paper | Choi, S. D. and Chang, Y. S.: Factors affecting the distribution of the rate of carbon uptake by forests in South Korea, Environ.Sci.

Table 2 .
Comparison of biomass carbon (C) density, biomass C stocks, and annual C balance rate of South Korean forests with those of two previous studies.

Table 3 .
The estimates of annual carbon (C) sink and net biome production (NBP) compared to those in past studies.

Table A1 .
Parameter estimates of the Gompertz function for stem volume (m 3 ha −1 ) for six dominant species by site index in Korea.Volume(age) = a • exp(b • exp(c • age)).