Chapter 20
Downloaded by NATL UNIV OF SINGAPORE on May 5, 2018 | https://pubs.acs.org Publication Date (Web): October 11, 2011 | doi: 10.1021/bk-2011-1072.ch020
Development of Spatial Inventory of Nitrous Oxide Emissions from Agricultural Land Uses in California Using Biogeochemical Modeling Lei Guo,*,1 Dongmin Luo,1 Changsheng Li,2 and Michael FitzGibbon1 1Research
Division, California Air Resources Board, Sacramento, California 2Institute for the Study of Earth, Oceans, and Space, University of New Hampshire, Durham, New Hampshire *E-mail:
[email protected] Nitrous oxide (N2O) is a potent greenhouse gas (GHG) that contributes to global warming. In California, agricultural soil management is recognized as an important source of N2O. It contributed over 50% of the total N2O inventory in 2008. We evaluated N2O emissions from agricultural soils in California using the biogeochemical model Denitrification-Decomposition (DNDC). Emission fluxes of N2O from 17 types of agricultural land uses in 49 counties were simulated based on California-specific data of soil, land use, meteorology, and land management practices. The study indicated that N2O fluxes varied tremendously across the landscape. Total annual N2O emission derived directly from all agricultural soils statewide was about 1.27 x 104 ton N, with more than 83% from the Central Valley. Annual fluxes of N2O ranged from 15.4 Kg N/ha from cotton fields to 0.03 Kg N/ha from rice paddies, with an average of 3.7 Kg N/ha from all agricultural land uses. Kings is the highest emitting county, which contributed approximately 17% of the total N2O emissions from California’s agricultural soils, followed by San Joaquin (11%), Fresno (11%), and Tulare (11%) counties. We provided emission maps displaying spatial distribution of N2O emissions from agricultural soils that reflected local site-specific conditions. However, our emission estimates are subject to significant uncertainties with regard to
© 2011 American Chemical Society Guo et al.; Understanding Greenhouse Gas Emissions from Agricultural Management ACS Symposium Series; American Chemical Society: Washington, DC, 2011.
Downloaded by NATL UNIV OF SINGAPORE on May 5, 2018 | https://pubs.acs.org Publication Date (Web): October 11, 2011 | doi: 10.1021/bk-2011-1072.ch020
the input data, especially farming management parameters, and should not be viewed as real case emission scenarios. The study nevertheless demonstrates the usefulness of process-based geochemical modeling in assessing spatial distribution of N2O emissions from agricultural soils. Extensive field studies are underway monitoring N2O fluxes from major California cropping systems. The results from these studies will be used to improve our model and thus reduce the uncertainties of the emission estimates.
Introduction Nitrous oxide, or N2O, is a naturally occurring gas with an estimated lifetime of 120 years in the atmosphere (1). With its broader absorption spectrum in the infrared range, N2O is one of the most potent naturally occurring greenhouse gases (GHGs). The Intergovernmental Panel on Climate Change (IPCC) has determined its Global Warming Potential (GWP) to be 298 CO2 equivalent (CO2e) over 100year time horizon, more than ten times that of the next significant contributing GHG methane (CH4) which has a GWP of 25 (2). N2O is mainly produced from natural processes of nitrification, denitrification, and combustion. Nitrification and denitrification are common soil processes which play a vital role in nitrogen (N) cycling in ecosystems. However, agricultural activities involving intensive soil management, such as N fertilizer application and irrigation, can enhance nitrification and denitrification, causing elevated N2O emissions that are far above the natural background. According to United States Environmental Protection Agency (USEPA), nitrification and denitrification are responsible for approximately 7.5 x 105 ton of N2O produced in the United States in 2008, primarily from agricultural soils associated with crop production (3). In California, about 2.3 x 104 ton of N2O was estimated to be emitted from agricultural soils in 2008, representing 54% of the total N2O inventory in the State (4). The conventional way of estimating N2O emissions from agricultural soils is to use the emission factor (EF) approach, which assumes that a fixed fraction of the nitrogen applied to the soil is converted to N2O. However, emission of N2O from soil is a microbe-driven process, affected by numerous environmental factors that govern microbial activities. Fluxes of N2O emissions are found to be related not only to N fertilizer application rate (5–7), but also to soil organic matter content (8–10), soil water content (11–14), soil pH (7, 15–17), land cover (6, 18, 19), management practices (20–23), as well as meteorological conditions (18, 19, 24–26). Due to both spatial and temporal variability of N2O fluxes, it is extremely challenging to characterize N2O emissions from agricultural soils quantitatively. Process-based biogeochemical models such as DAYCENT (27–30) and DNDC (31–33) have been developed to characterize the complicated interactions of biological, chemical, and physical processes in soil, and used to simulate emissions of trace gases produced from these interactions. USEPA has employed 388 Guo et al.; Understanding Greenhouse Gas Emissions from Agricultural Management ACS Symposium Series; American Chemical Society: Washington, DC, 2011.
Downloaded by NATL UNIV OF SINGAPORE on May 5, 2018 | https://pubs.acs.org Publication Date (Web): October 11, 2011 | doi: 10.1021/bk-2011-1072.ch020
a hybrid methodology combining both DAYCENT modeling and the IPCC emission factor approach in the development of national N2O inventory from agricultural soils (3). The purpose of this study was to explore and demonstrate the use of the biogeochemical model DNDC as a methodology to estimate N2O emissions from various cropping systems in California and to develop N2O emission maps that reflect site specific crop, soil, weather, and agricultural land management conditions. DNDC has been applied and verified for many cropping systems worldwide (34–37). The model has also been used to assess carbon dynamics and sequestration potential of agricultural soils in California (38).
Methodology DNDC Model Denitrification-Decomposition, or DNDC, is a biogeochemical model formulated to simulate carbon (C) and N interactions and cycling in agricultural ecosystems (31–33). Built upon fundamental biogeochemical processes of decomposition, fermentation, nitrification, and denitrification, the DNDC model incorporates classic laws of physics, chemistry, and biology as well as many empirical equations developed from extensive scientific literatures. The model is capable of predicting dynamics of carbon and nitrogen species, including production of trace gases of CO2, CH4, N2O, NOX, and NH3 in ecosystems, based on the basic ecological drivers of crop, soil, weather, and management activities (Figure 1). It consists of three submodels which simulate the mass transfer of heat and water (the thermal-hydraulic submodel), carbon species (the decomposition submodel ), and nitrogen species (the denitrification submodel), respectively. DNDC can be used to analyze C and N cycling at the field, regional, or national scale depending on the spatial resolution of the input GIS database that specifies temporal and spatial variations of the basic ecological drivers.
Figure 1. DNDC data requirements and outputs. 389 Guo et al.; Understanding Greenhouse Gas Emissions from Agricultural Management ACS Symposium Series; American Chemical Society: Washington, DC, 2011.
Downloaded by NATL UNIV OF SINGAPORE on May 5, 2018 | https://pubs.acs.org Publication Date (Web): October 11, 2011 | doi: 10.1021/bk-2011-1072.ch020
Data Sources Most of our data sources were from the public domain. The land use data was obtained from the Land Use Survey of the California Department of Water Resources (CDWR) (39) and the 2002 Census of Agriculture of the National Agricultural Statistics Service, the U.S. Department of Agriculture (USDA) (40). The CDWR’s Land Use Survey data for the counties simulated in this study was collected between mid-1990s and 2006 and contains spatial information on agricultural, urban, and native vegetation lands, including description of land cover, acreage, and, when appropriate, irrigation method and water sources. The urban and native vegetation lands were eliminated, however, from the simulation because our study focused only on agricultural lands. The USDA’s Census of Agriculture provides only agricultural land use data, aggregated at the county level. We simulated GHG emissions from 49 out of the 58 counties in California, covering an area of 3.4 x 106 ha, about 99% of the total harvested crop lands (referred to as “agricultural land uses” thereafter) in the State. The N2O emissions were simulated using the data from the 2002 Census of Agriculture, but spatial distribution of the N2O fluxes was allocated based on the CDWR’s Land Use Survey data. Seventeen types of agricultural land uses were simulated as shown in Table 1. We did not model land uses at the subclass level of the CDWR’s Land Use Survey, which tracked down specific crops. Instead, all crops were grouped into broad categories based on their phenological and physiological characteristics. For example, eggplants, peppers, cucumbers, and tomatoes, etc. were all grouped as vegetables. The tree crops were divided into either deciduous or evergreen orchards. The soils data source was from the Soil Survey Geographic database (SSURGO) of the Natural Resources Conservation Service, USDA (41). SSURGO is a complicated soil survey database containing detailed soil distribution and profile information for the entire United States. Four soil properties were taken directly from SSURGO: soil organic carbon (SOC) content, soil density, soil pH, and soil clay fraction. We calculated the area-weighted means of the four soil properties of the top soil horizons for each county by overlying the polygons of soil components layer (COMP) from SSURGO in that county with those of the agricultural land uses from the Land Use Survey of the CDWR. These four soil parameters were then used as basic drivers to establish, based on the built-in empirical relations of the DNDC model, other soil characteristics, such as soil porosity, saturated hydraulic conductivity, field capacity, wilting point, and specific heat, required for the DNDC model. The meteorological data was obtained from the California Weather Database (42) of the University of California (UC), Davis. The UC database stores current and historical weather data for approximately 400 weather stations throughout California. Daily precipitation, minimum and maximum temperatures, and solar radiation data are available from three network sources: (1) the California Irrigation Meteorological Information System (CIMIS) stations of the CDWR, (2) the National Oceanic and Atmospheric Administration (NOAA) stations of the United States Department of Commerce, and (3) the TouchTone (TT) stations of the UC TT Network. The data records of all networks were pooled to obtain 390 Guo et al.; Understanding Greenhouse Gas Emissions from Agricultural Management ACS Symposium Series; American Chemical Society: Washington, DC, 2011.
Downloaded by NATL UNIV OF SINGAPORE on May 5, 2018 | https://pubs.acs.org Publication Date (Web): October 11, 2011 | doi: 10.1021/bk-2011-1072.ch020
a coverage as complete and uniform as possible in the simulated area. Missing records were filled by taking averages of nearby stations. We used “typical” management practices in the simulation. The management practices were developed largely from the University of California Cooperative Extension (UCCE) reports; the Cost and Return Studies of the Department of Agricultural and Resource Economics, UC Davis; and personal communications with the UCCE staff. The irrigation practices were, however, simulated using the DNDC Irrigation Index 1, which sets soil water content automatically to field capacity to meet plant demands at 100%. The option of Irrigation Index 1 represents optimum irrigation conditions where no over-irrigation or water stress occurs. In reality, however, irrigation methods in California are extremely diversified, covering practices from broad furrow flooding to high precision micro-sprinklers or subsurface dripping. Irrigation records are one of the data sources that are especially difficult to obtain, or do not exist.
Model Scenarios Our model scenarios were developed to represent baseline N2O emissions in 49 out of 58 California counties. Land uses in the nine remaining counties are either dominantly urban or native vegetation. The DNDC model was run using the meteorological data of 1990 to 2008 for each county. The model was first calibrated for each of the 17 major crop types to achieve a net carbon, nitrogen, and heat balance in and out of the cropping system while sustaining the expected yield and total biological mass produced. The final calibrated model thus represented the overall mass flow and plant growth well, although not validated against sitespecific N2O emission fluxes due to lack of monitoring data. To simulate baseline N2O emissions, we used “typical” management practices and “representative” soil parameter values. No alternative scenarios with regard to management activities were simulated that would produce different emission estimates of N2O. Soil emissions with minimum and maximum soil parameter values for the four basic soil properties, i.e., soil organic carbon content, soil density, soil clay content, and pH, were also simulated for uncertainty analyses. These estimates would provide potential range of emission estimates due to variability of soil properties. Additional uncertainties associated with management practices such as irrigation and crop residue management were not explored in the study. Four nitrogen sources were included in the modeling study: (1) N fertilizer application; (2) biological fixation of atmospheric N2 in legume and, to a lesser extent, non-legume crops; (3) land application of livestock (primarily dairy) wastes; and (4) atmospheric deposition of nitrogen from precipitation. Application of livestock wastes or manure was only made in the eight counties where significant dairy operations exist: Fresno, Kern, Kings, Madera, Merced, San Joaquin, Stanislaus, and Tulare. About 28% of the total N generated from dairy manure in these counties was assumed to be applied to the following forage crops: corn, wheat, oats, sorghum, alfalfa, and non-legume hay (R. Zhang, personal communication; (43, 44)). The application of chemical nitrogen 391 Guo et al.; Understanding Greenhouse Gas Emissions from Agricultural Management ACS Symposium Series; American Chemical Society: Washington, DC, 2011.
fertilizers in those crops was reduced accordingly to maintain the same nitrogen rates as in other counties without manure applications.
Downloaded by NATL UNIV OF SINGAPORE on May 5, 2018 | https://pubs.acs.org Publication Date (Web): October 11, 2011 | doi: 10.1021/bk-2011-1072.ch020
Results and Discussion As shown in Figure 2, about 9% of California land, or 3.5 x 106 ha, is used actively for agricultural production, most of which is located in the Central Valley. Major agricultural counties include Fresno, Kern, Tulare, Kings, and San Joaquin in the San Joaquin Valley, and Glenn, Colusa, Yolo, Butte, and Sutter in the Sacramento Valley. According to the 2002 Census of Agriculture (40), major crops cultivated in California are orchards (1.2 x 106 ha), forage hay (7.9 x 105 ha), vegetables (4.8 x 105 ha), cotton (2.8 x 105 ha), and corn (2.3 x 105 ha).
Figure 2. Distribution of agricultural land uses in California. (Data source: Land Use Survey, California Department of Water Resources. Data retrieved March, 2009.) (see color insert)
392 Guo et al.; Understanding Greenhouse Gas Emissions from Agricultural Management ACS Symposium Series; American Chemical Society: Washington, DC, 2011.
Table 1. List of agricultural land uses, cultivation areas, and N fertilizer application rates simulated in the DNDC modeling.
Downloaded by NATL UNIV OF SINGAPORE on May 5, 2018 | https://pubs.acs.org Publication Date (Web): October 11, 2011 | doi: 10.1021/bk-2011-1072.ch020
Area
Fertilizer rate
Crop
ha
% total
Kg N/ha
Alfalfa
444510
13.05
18
Barley
28903
0.85
45
Beans
22923
0.67
88
Corn
226290
6.64
255
Cotton
280242
8.23
280
Deciduous orchards
608689
17.87
112
Evergreen orchards
179228
5.26
123
Non-legume hay
333413
9.79
224
Oats
12801
0.38
67
Potato
21489
0.63
305
Rice
214726
6.30
135
Sorghum
4664
0.14
157
Sugarbeets
22373
0.66
210
Sunflower
6464
0.19
105
Vegetables
475130
13.95
184
Vineyards
359497
10.56
50
Wheat
164320
4.82
144
Total
3405662
100
Table 2 lists the N2O fluxes, emission factors (i.e., emission potentials), and total N2O emissions for the 17 types of agricultural land uses simulated in this study. Although we ran the DNDC model using historical meteorological data, we used static land use data (most closely represented by year 2002), and constant crop management data in the simulation. As shown in Table 2, total statewide annual emission of N2O derived directly from all agricultural land uses was estimated to be 1.27 x 104 ton N, equivalent to 6.17 million metric tons (mmt) CO2e. Cotton and non-legume hay contributed almost 60% of the total N2O emission, followed by alfalfa (14%), corn (10%), vegetables (9%), and deciduous orchards (3%). The least contributing crop was rice, whose emissions constituted < 0.05% of the total N2O. The relative contribution of a particular land use to the total N2O emission is dependent on both its emission fluxes and its cultivation area. For example, potato system generates high N2O fluxes, but its contribution to the total N2O emission in California was only