It is widely known that greenhouse gas (GHG) emissions contribute to global climate change. Soil respiration is known to be one of the largest CO2 fluxes to the atmosphere. In the Korean Peninsula, rice is generally cultivated in rice paddy fields, whose flooding conditions create anaerobic systems below the soil surface. Due to these conditions of rice paddy fields, these fields are one of the major sources of methane (CH4 ) emissions to the atmosphere. However, rice paddy ecosystems are not adequately represented in most Land Surface Models (LSMs). Most LSMs are limited in predicting GHG effluxes from the soil surface. The objectives of this study are to link an LSM with biogeochemistry models, and to estimate the land surface states and fluxes including evapotranspiration, soil temperature, and CO2 and CH4 effluxes from various types of landcover, including rice paddy fields, for the Korean Peninsula, by applying the tool to the region.

Agroclimatic Zones (ACZs) were initially classified into 19 zones for rice cultivation. However, it is difficult to collect all the required information for each of the 19 ACZs to estimate CO2 and CH4 effluxes from the soil surface using biogeochemistry models. Therefore, for this study, we designated 9 new agro climatic zones which were classified ACZ1 to ACZ9. Estimation of soil respiration for urban regions was not included in this study. Representative cells for each landuse type, including paddy fields, upland, and forest, in each ACZ were selected based on land use and soil characteristics. This selection was performed using vector types of land use and detailed soil maps. Additionally, we checked if the land use of selected representative cells at a 5-km resolution on a raster type of land use map were the same as on the vector type of land use map, since NOAH-LSM was simulated with a 5 km resolution. Because only 39 grid cells with a 5 km resolution were identified and this number of upland cells was much less than the 2,985 forest cells and 503 paddy field cells, the upland cells were excluded for the GHGs estimation. Even though we initially tried to use observed meteorological datasets from the Automated Surface Observing System (ASOS), due to the relatively long distances between the representative cells and the nearest ASOSs, Tropical Rainfall Measuring Mission (TRMM) 3B42 and NCEP Global Data Assimilation System (GDAS) were used for this study instead.

NOAH-LSM has been developed by a number of groups, including the NCEP Environmental Modeling Center (NCEP/EMC), Oregon State University (OSU), the NWS Hydrology Lab (HRL), the Air Force Weather Agency (AFWA), and the Air Force Research Lab (AFRL). It is able to simulate land surface states and fluxes, including soil moisture (both liquid and frozen), soil temperature, skin temperature, snowpack depth, snowpack water equivalent, canopy water content, evapotranspiration, and runoff and has been operationally used in NCEP models since 1996. For this study, NOAH-LSM (version 3.2), a 1-dimensional column model, was used to provide evapotranspiration and soil temperatures at multiple depths below the soil surface for the SOILCO2. The National Centers for Environmental Prediction Global Data Assimilation System (NCEP GDAS) forcing dataset was used to drive the simulations. The precipitation field, taken from the Tropical Rainfall Measuring Mission (TRMM) 3B42 dataset, was used for supplemental forcing sources. The U.S. Geological Survey (USGS) land cover dataset, The United Nations' Food and Agriculture Organization (FAO) soil data for sand-, clay-, and silt-fraction maps, and GTOPO30 were also used for the NOAH-LSM simulation. NOAH-LSM was simulated with a 5-km resolution and 1-hr time step size. We checked if the resolutions of the landcover, soil, and Digital Elevation Model (DEM) were sufficiently high for the simulation at 5-km resolution. We found that the FAO soil fractions were too coarse for the simulation. The FAO soil fractions were replaced with a detailed soil map of Korea for the simulation domain (33-38.5° N and 125-130° E).

Numerical modeling approaches were used to estimate CO2 and CH4 effluxes from the soil surface. The SOILCO2 model, a one-dimensional process-based model, is one of the components of the HYDRUS-1D software package. HYDRUS-1D simulates water flow, heat, solute, and CO2 transport. The Denitrification-Decomposition (DNDC) model is a process-based model to simulate carbon and nitrogen dynamics of agricultural fields, including rice paddy fields. Agricultural management specifications, including fertilization and irrigation, are used as inputs for the DNDC model, as well as meteorological datasets, such as daily precipitation and air temperature. The SOILCO2 model has been used to estimate CO2 efflux from the soil surface and carbon production in soils. The SOILCO2 and DNDC models were applied to estimate CO2 effluxes from forests and CH4 effluxes from paddy fields. The DNDC model was used to estimate CH4 effluxes from paddy fields, since the DNDC model considers flooding conditions and various agricultural management techniques. For the simulation of the SOILCO2 model, evaporation, transpiration, and soil temperature were used. However, these datasets were not easy to observe. Furthermore, evapotransipration is difficult to divide into two separate variables: evaporation and transpiration. The daily evaporation and transpiration and soil temperature resulting from the NOAH-LSM simulations were fed into the SOILCO2 model. We applied this linked model framework to the representative cells in each ACZ and estimated the annual CO2 and CH4 emissions (kg C yr-1 ) from paddy fields and forest at the “gun”-level by multiplying CO2 and CH4 effluxes (kg C ha-1 ) by area of paddy fields and forest, respectively. The observed data from the Gimje flux site were used to evaluate these models’ performance.

The daily net radiation observed at the Gimje flux tower site was used to evaluate NOAH-LSM. 3-hour interval outputs from NOAH-LSM were aggregated to produce daily and monthly outputs and daily net radiation (the difference between daily net longwave radiation and net shortwave radiation), which was compared with the observed daily net radiation. The residual of the daily net radiation in winter was somewhat higher than in summer. NOAH-LSM simulated daily net radiation with 28.82 Wm-2 RMSE. Monthly mean evapotranspiration in July and August 2012 was higher than that in June and September 2012 over the country. Monthly mean soil temperature at the first layer ranged from 259.5 to 304.9 K and at the fourth layer it ranged from 271.5 to 300.7 K. These values were used to simulate CO2 effluxes from forests using the SOILCO2 model. The range of CO2 effluxes from forests in ACZ2 was from 4.27 to 134.97 kg CO2 ha-1 d-1 . The estimated annual cumulative CO2 emissions from forests at the “gun”-level ranged from approximately 2.7 × 106 to 113.7 Ton CO2 ha-1 . The maximum annual cumulative CO2 emissions were estimated at Hongchun-gun in Ganwond-do and the minimum annual cumulative CO2 emissions were estimated at Dong-gu in Incheon. The DNDC model was used to estimate CH4 effluxes from rice paddy fields. The model results were evaluated with the observed CH4 effluxes from the Gimje flux tower site. The RMSE for daily CH4 effluxes from transplanting to harvesting in 2012 was 1.4 kg C ha-1. The estimated annual cumulative CH4 emissions from rice paddy fields at the “gun”-level showed that Haenam-gun in Jeonnam-do was the largest CH4 emitting “gun” and Gurae-gun in Jeonnam-do was the smallest CH4 emitting “gun”, not counting urban regions.

The conclusions of this study are as follows:

1. Since agricultural management techniques, including fertilization and irrigation, are used to estimate CH4 emissions from rice paddy fields using the DNDC model, it is concluded that the DNDC model can be used for effective adaptation plans for the reduction of CH4 emissions from rice paddy fields in Korea.

2. This result suggests that long-term monitoring of GHG emissions contributes to the improvement and evaluation of biogeochemistry models, such as DNDC.

3. It is concluded that since soil respiration is one of the major factors for estimating Net Ecosystem Exchange (NEE), this study can be useful for understanding whether the forest ecosystem is a source or sink of carbon.

4. We hope that this study can be used to provide effective GHG reduction plans by revealing the local characteristics of GHG emissions by providing GHG emissions at the “gun”-level.

1. 서론

2. 연구 자료 및 방법

2.1 자료수집

2.2 LSM-Biogeochemistry 연계모형

3. 연구결과 및 토의

3.1 LSM-Biogeochemistry 연계모형

4. 요약 및 결론