Land surface model with vegetation dynamics

The interactions between land and atmosphere are essential features to describe the function of terrestrial ecosystems in climate regulation. These interactions are typically implemented as a module in climate and weather forecasting models at the global or regional scale. The first incorporation of land surface processes in a global or regional atmospheric model was a simple bucket model that considered only a one-layer surface (Manabe, 1969). Then, improved LSMs were developed, which considered more complex processes in the energy and water cycles (Deardorff, 1978; Dickinson et al., 1986; Eagleson, 1982; Pan & Mahrt, 1987). Since terrestrial ecosystems have been known to play very important roles in the energy and water cycles, climate models have increasingly introduced vegetation schemes to reflect the physiological and phenological processes of vegetated surfaces. To quantify the emissions of greenhouse gases, it is important to represent the biological mechanism for the carbon cycle within vegetated surfaces via the photosynthetic activities and respiration of vegetation. In the past 20 years, dynamic vegetation models have been developed to simulate the temporal variations in biogeophysical and biogeochemical processes with those in atmospheric features such as temperature, PRCP, wind, and etc. In addition to considering the interactions between land and atmosphere, these models also embody functional relationships between the climate and ecological factors (Daly et al., 2000; Prentice et al., 2000; Running & Couglan, 1988).

The Noah LSM with multi-parameterization options (Noah-MP) was introduced to represent the changes in climate regulation function of terrestrial ecosystems. Noah-MP is an advanced version of the Noah LSM (Niu et al., 2011). The Noah LSM describes land and atmospheric interactions based on thermal energy and water balance and was first developed by Oregon State University (Pan & Mahrt, 1987). This model was extended by incorporating a vegetation transpiration scheme and a surface runoff scheme (Chen et al., 1996). Since then, this model has been widely used in a standalone off-line mode, as well as in a coupled mode with a regional weather forecasting model (Chen & Dudhia, 2001). On the basis of the Noah LSM 3.0, Noah-MP provides additional scheme options: a dynamic vegetation model (Dickinson et al., 1998), a simple groundwater model with a TOPMODEL runoff scheme (Niu et al., 2005), a radiation transfer scheme modified for the three-dimensional structure effect of vegetation canopy (Niu & Yang, 2004), a 3-layer snow model (Yang & Niu, 2003), and a frozen soil scheme for separated permeable and impermeable fractions (Niu & Yang, 2006). Thus, Noah-MP allows a huge number of ensemble simulations to be conducted via scheme combinations. Through various experiments using Noah-MP, Yang et al. (2011) proposed a combination of physical schemes for more accurate simulations at the global and continental scales. Hong et al. (2014) investigated the applicability of genetic algorithms to determine an optimal set of physical schemes for Noah-MP for South Korea, and Hong et al. (2015) successfully applied this approach to East Asia. In this study, the optimal scheme combination of Noah-MP determined by Hong et al. (2015) was used to extract the variables related to the climate regulation function of vegetated surfaces over East Asia.

Data and study domain

To operate Noah-MP, we obtained atmospheric forcing data from the Global Land Data Assimilation System (GLDAS) developed by NASA (Rodell et al., 2004). This data comprises reanalysis data obtained by assimilating atmospheric data from various operational forecasts and observations, for example, from data from the Global Data Assimilation System National Oceanic and Atmospheric Administration of USA. The list of six inputs as atmospheric forcings from GLDAS and outputs as variables in land surface processes is shown in Table 1. The temporal and spatial resolution of the final Noah-MP outputs are 3-hour and 0.25° (approximately 25 km), which are the same as those of GLDAS atmospheric forcing data. The GLDAS data from 1960 onwards are available at ldas.gsfc.nasa.gov.

This model requires land cover and soil texture information over the study area as static inputs. Land cover information, in particular, is used to characterize the biogeochemical features over vegetated covers. Land cover information was from the land use and land cover data used by Kang et al. (2010) (hereafter land cover data from Kongu national university [KLC]), which were generated from the normalized difference vegetation index data of the Moderate Resolution Image Spectro-radiometer (Fig. 1). The land cover classification method used in KLC follows that of the International Geosphere-Biosphere Programme (Friedl et al., 2002). The soil texture data that follow a typical triangular grain-size-based classification were introduced from the Food and Agricultural Organization (FAO, 2002). The study domain was set from −10 to 90° and from 90 to 150° E in latitude and longitude, respectively.

The Noah-MP simulations begin with arbitrary initial conditions regarding the surface status, such as initial soil moisture, SH, LH, and water table depth. Thus, a spin-up process, which is a warm-up simulation for a certain period, is required for the surface conditions to reach equilibrium. With the exception of water table depth, the surface conditions in Noah-MP typically reach equilibrium after an approximately 10-year spin-up (Cai et al., 2014). Thus, in this study, a 60-year simulation was performed from 1951 to 2010 and only 50-year model outputs from 1961 was used for the analyses. The outputs from the initial 10-year period were excluded as a spin-up.

Quantification of climate regulation

Three main components can be considered to quantify the contributions of the surface to climate change. The first is the effect of surface fluxes related to greenhouse gases (ex. CO₂, NO₂, CH₄, etc) absorbed or emitted through biogeochemical processes at the surface. A larger amount of greenhouse gas absorbed by the ecosystem leads to higher functional improvement of the climate regulation. Of the greenhouse gases mentioned above, carbon fluxes at the terrestrial ecosystem can be presented by the Net Ecosystem Exchange (NEE). NEE represents the net amount of emitted or absorbed carbon per unit area and time at the surface. NEE can be converted into radiative forcing for CO₂ as follows (Anderson-Teixeira & DeLucia, 2011).

G_CO2 is the equivalent radiation forcing of CO₂ gas at a given time step, and E is the effective radiative efficiency of CO₂ (1.4×10⁴ nWm^-2ppb^-1) (Anderson-Teixeira & DeLucia, 2011; Forster et al., 2007). C(t) represents the additional CO₂ concentration in the atmosphere as a rate of the CO₂ moles of emitted from the ecosystem surface to the atmosphere. D denotes the CO₂ pulse decay over time. In this study, the annual variations in climate regulation were investigated, and thus D becomes about 0.875 (k mol ha^-1 yr^-1). E is the effective radiative efficiency of CO₂ (1.4×10⁴) (Foster et al., 2007).

The second component is net radiation (R_n), which represents the net amount of incoming and outgoing thermal energy fluxes at the surface. This component is a biogeophysical contributor. R_n is determined mainly by albedo, the radiation reflectivity of the surface. Since albedo is determined by the characteristics of vegetation covers, terrestrial ecosystems play an important role in the energy exchange. R_n is quantified by the net sum of incoming and outgoing radiation that are balanced with transferred energy forms such as SH, LH, and GH. The variance of R_n (∆R_n) during a given time period represents the total residual amount in the atmosphere due to energy transfer processes at the surface, which can be categorized as positive or negative feedback. Thus, a higher ∆R_n value indicates a greater degradation of the climate regulation function.

LH can be considered as the third contributor to the climate regulation function. It represents the surface energy flux due to the evapotranspiring process that includes direct evaporation at soil surfaces and transpiration from vegetation. Transpiration is accompanied with photosynthetic activities. Since the water vapors emitted to the atmosphere via this process are resupplied to the land surface and lead to cooling effects, LH can be considered as a positive effect in terms of the climate regulation function of terrestrial ecosystems.

Anderson-Texieira et al. (2012) proposed an equation to quantitatively represent the contribution of terrestrial ecosystems to climate change by combining the three aforementioned components as follows.

Positive values of ∆F_CR, the variance of climate regulation function of terrestrial ecosystems, indicate that the ecosystem is functioning to regulate the increase in global warming. The quantitative values of ∆F_CR can be numerically calculated when all three components are in the same unit as radiative flux (Wm^-2).

General trends of climate change over East Asia

Fig. 2 describes the spatio-temporal climate-change trends over East Asia which was from the temperature and PRCP data used for the Noah-MP forcing inputs. Over 50 years, most East Asia regions have exhibited temperature increases, but higher latitudes have shown higher warming trends; regions near the Arctic have shown temperature increases up to 0.05°C per year (Fig. 2A). The temperature variabilities, however, have decreased at high latitudes, indicating that the rates of increase of the annual minimum temperatures are higher than those of the maximum temperatures (Fig. 2B). It is assumed that these increasing trends but decreasing variabilities in temperature at high latitudes are due to higher temperature increases during winter and spring than those during summer and fall (not shown), which reduces inter-seasonal differences in temperature.

Variations in PRCP have been relatively high at 30 to 50° latitudes for 50 years (Fig. 2C): decreasing trends in northern part of China and southern part of Japan and increasing trends in southern part of the Gobi Desert and Korean Peninsula. The variabilities in PRCP, on the contrary, have increased in the Korean Peninsula and southern part of Japan (Fig. 2D). Interestingly, considering the contrasting PRCP trends over these two regions, the PRCP regimes have progressed very differently from each other. For example, the PRCP information can indicate the increasing possibility of extreme events such as heavy rainfall and severe drought in the Korean Peninsula and Japan, respectively. This indication of the variations in extreme events over the two regions must be verified further with more PRCP information at higher spatial and temporal resolutions. The variational trends over the tropics, however, are not significant, considering the very high averages in annual total PRCP.

Variations in quantified climate regulation from terrestrial ecosystem over East Asia

The spatial distributions of the 50-year trends in the quantified climate regulation due to terrestrial ecosystems over East Asia is shown in Fig. 3A. The final outputs were obtained from equation 2 after the quantification of each component simulated using the Noah-MP model. The regions showing an increasing trend of F_CR are high latitude regions around Siberia, the Korean Peninsula, Japan Islands, and southeast part of China around the Yangzi River. Meanwhile, those showing a decreasing trend are mainly in the tropics at low latitudes. Interestingly, around the Gobi Desert, the F_CR decreases eastward but increases southward.

The significance of the variation in F_CR can be examined by calculating it as a percentage of the 50-year average for each region (Fig. 3B). For instance, the increasing trends over Siberian regions are significant, showing increases up to 0.4% per year, which arithmetically indicates an increase over 20% for the past 50 years. Unfortunately, the embedded biogeochemical processes in Noah-MP cannot represent possible offsetting effects from greenhouse gas fluxes via the warming and melting processes of permafrost such as methane. This may result in an overestimation in F_CR over such regions. Nevertheless, assuming that the atmospheric variations used for the model inputs are reflected by those in terrestrial ecosystems with climate change, the ecological changes and their impact on the atmosphere have been apparent over the regions. In addition, the differences in the trends F_CR between the eastern and southern parts of the Gobi Desert clearly imply the eastward expansion and southward shrinkage of desert areas. This is due to LH effect caused by change in precipitation regimes in the regions. However, it is unclear whether the variations reflect ecological changes over the regions due to very small magnitudes of F_CR. The tropics showed high decreasing trends in F_CR, but their impacts may not have been significant, comparing to the ordinary fluctuation scales of climate regulation function.

The 50-year trends of the three components, G_CO2, LH, and R_n, are shown in Fig. 4. The distribution pattern of G_CO2 is very similar to that of the total F_CR. This implies that the enhancement in F_CR over Siberian regions resulted mainly from biogeochemical processes. Note that some areas in Fig. 4A are blank because some vegetation schemes, such as desert land covers like the Gobi Desert, were not activated in Noah-MP. The contributions from LH to F_CR have been relatively low in Siberian regions (Fig. 4B). On the contrary, regions in the Korean Peninsula and around the border between China and Russia around the southeastern part of Siberia show increases in LH. Contributions of variations in net radiation derived mainly from biogeophysical processes have been relatively low over the entire East Asia region (Fig. 4C). Note that the opposite direction of color bar due to the opposite feedback to the atmosphere as the definition of climate regulation mentioned above. Thus, the contributions to the total F_CR from the three components can be arranged as follows, in decreasing order: G_CO2, LH, and R_n. In other words, warming over East Asia has led to biogeochemical processes at the land surfaces at high latitudes.

The averaged 50-year trend of F_CR for each land cover type is shown in Fig. 5. The trends of G_CO2 and LH reflect the regional characteristics of terrestrial ecosystems. Overall, majority of the land covers showed increasing trends of G_CO2 and LH, except evergreen broadleaf forests that are mainly distributed in the tropics. “Mixed shrub/grassian,” and “deciduous needleleaf forest” that are the major types of land cover in the Siberian regions showed increases in both components. The opposite phenomena over the “grassian” type of land cover that is mainly distributed around Gobi Desert is also related to the changes in the desert characteristics as mentioned above. Variations in R_n for land cover types have been relatively small, but most of them showed increasing trends, which has negatively affected the total F_CR.

In conclusion, Spatio-temporal variations in F_CR and its components, G_CO2, LH, and R_n, at the land surface were investigated to see their contributions to the past global warming. Those components were forced by atmospheric changes over East Asia in a LSM that represents the interactions between terrestrial ecosystems and the atmosphere. Atmospheric changes according to GLDAS data indicated that the warming trends during the last half century have been stronger at higher latitudes. Strong climatic signals in PRCP variations were also shown in regions between 30° and 50° latitudes. The simulated F_CR forced by these atmospheric variations showed significant increasing trends in land covers over high latitudes, and the main contributor was G_CO2 via biogeochemical processes. These increases in F_CR over the regions suggest the possibility of negative feedback to global warming accompanied by surface ecological changes. In other words, it can be concluded that global warming has stimulated biogeochemical processes at high latitudes, but the terrestrial ecosystems have worked to decrease the warming rate by increasing carbon assimilation. The model also described the variations in F_CR around the Gobi Desert, reflecting the possible eastward movement of desert areas.

These results showed that the F_CR simulation is sensitive to climate change and the characteristics of land cover types. This indicates that the accuracy of the model can be improved by using accurate information of spatial distributions of land covers, accurate parameterizations representing biogeochemical processes, and application of a long term land cover change model. The current model used in this study cannot reflect ecological successions and spatial changes in climate zones due to temporally fixed land covers. Another limitation of the LSM used in this study is that it cannot represent the possible effects of melted permafrost over high latitudes. Since this study considered only feedback from CO₂ gas involved in respiration at the surface, the increase in F_CR over high latitudes may be overestimated. The characterization of microbial processes involved in carbon fluxes at the soil surface must be improved.

This study demonstrated how the climate regulation function of terrestrial ecosystems has regionally changed with climate change for the last half century. Climate regulation due to terrestrial ecosystem varies with climatic factors in the atmosphere, and thus better understanding of the feedback from the terrestrial ecosystem to the atmosphere helps develop climate prediction models with higher accuracy. In addition, since regional characteristics of terrestrial ecosystems determine the climate regulation function, the approach used in this study can provide useful information to devise countermeasures for climate change mitigation at the regional scale. The information can be extracted by checking out the regional differences and variations in climate regulation components between the past 50 years and more recent ones in East Asia. For example, predicting changes in climate regulation due to future climate change can be valuable in creating land use policies to improve the climate regulation function and also serve as strategic information for international communities to mitigate climate change.