About Us | Help Videos | Contact Us | Subscriptions

Journal of Environmental Quality - Article



This article in JEQ

  1. Vol. 42 No. 4, p. 1100-1108
    unlockOPEN ACCESS
    Received: Dec 17, 2012
    Published: June 24, 2014

    * Corresponding author(s): Yaling.Qian@colostate.edu
Request Permissions


Simulation of Nitrous Oxide Emissions and Estimation of Global Warming Potential in Turfgrass Systems Using the DAYCENT Model

  1. Yao Zhanga,
  2. Yaling Qian *a,
  3. Dale J. Bremerb and
  4. Jason P. Kayec
  1. a Dep. of Horticulture and Landscape Architecture, Colorado State Univ., Fort Collins, CO 80523
    b Dep. of Horticulture, Forestry & Recreation Resources, Kansas State Univ., Manhattan, KS 66506
    c Dep. of Crop and Soil Sciences, Pennsylvania State Univ., University Park, PA 16802


Nitrous oxide (N2O) emissions are an important component of the greenhouse gas budget for turfgrasses. To estimate N2O emissions and global warming potential, the DAYCENT ecosystem model was parameterized and applied to turfgrass ecosystems. The annual cumulative N2O emissions predicted by the DAYCENT model were close to the measured emission rates of Kentucky bluegrass (Poa pratensis L.) sites in Colorado (within 16% of the observed values). For the perennial ryegrass (Lolium perenne L.) site in Kansas, the DAYCENT model initially overestimated the N2O emissions for all treatments (urea and ammonium sulfate at 250 kg N ha−1 yr−1 and urea at 50 kg N ha−1 yr−1) by about 200%. After including the effect of biological nitrification inhibition in the root exudate of perennial ryegrass, the DAYCENT model correctly simulated the N2O emissions for all treatments (within 8% of the observed values). After calibration and validation, the DAYCENT model was used to simulate N2O emissions and carbon sequestration of a Kentucky bluegrass lawn under a series of management regimes. The model simulation suggested that gradually reducing fertilization as the lawn ages from 0 to 50 yr would significantly reduce long-term N2O emissions by approximately 40% when compared with applying N at a constant rate of 150 kg N ha−1 yr−1. Our simulation indicates that a Kentucky bluegrass lawn in Colorado could change from a sink to a weak source of greenhouse gas emissions 20 to 30 yr after establishment.


    BMPs, best management practices; BNI, biological nitrification inhibition; GHG, greenhouse house gas; GWP, global warming potential; PET, potential evapotranspiration; SOC, soil organic carbon; SOM, soil organic matter; WFPS, water-filled pore space

Climate change is predicted to continue in the next several decades due to increases in anthropogenic greenhouse gas (GHG) emissions (IPCC, 2007). Managed ecosystems can serve as a GHG sink or source. Turfgrass, which is a highly managed ecosystem, occupies large areas of urbanized land. In the continental United States, turfgrass area is estimated to be three times larger than that of any irrigated crop (Milesi et al., 2005). Urban areas are expected to expand rapidly in the next a few decades, and the role of turfgrass management in the global GHG budget is not clear (Alig et al., 2004; Kaye et al., 2004; Townsend-Small and Czimczik, 2010).

Fertilizer application, irrigation, and other turfgrass management practices have the potential to contribute to emissions and mitigation of GHGs, leading to uncertainties in the net contribution of turfgrass ecosystems to climate change. Few studies have calculated the net global warming potential (GWP) for turfgrass ecosystems (Bartlett and James, 2011; Zirkle et al., 2011). The major components of GWP calculation for turfgrass ecosystems include (i) soil fluxes of nitrous oxide (N2O) and methane (CH4) (Bremer, 2006; Groffman et al., 2009; Groffman and Pouyat, 2009); (ii) GHG emissions associated with turfgrass maintenance, such as manufacturing and transporting fertilizer and pesticides, electricity used for irrigation, and fossil fuel combustion from mowers; and (iii) soil carbon (C) sequestration from the long-term storage of soil organic C (SOC) (Qian et al., 2003).

Fertilization of turfgrass has been shown to increase soil N2O emissions. The recorded emissions from fertilized turfgrass systems ranged from 0.5 to 6.4 kg N ha−1 yr−1 (Guilbault and Matthias, 1998; Kaye et al., 2004; Bremer, 2006; Groffman et al., 2009; Livesley et al., 2010; Townsend-Small and Czimczik, 2010). Soil N2O fluxes are mediated by microorganisms through nitrification and denitrification and are modulated by a suite of environmental factors, including oxygen, nitrogen (N), and C availability (Conrad, 1996). Denitrification from turfgrass was significant when soil was saturated (Mancino et al., 1988); in unsaturated or aerated soils, the majority of N2O emissions are expected to be from nitrification. Some plant species suppress nitrification by releasing nitrification inhibitory compounds from their roots as a mechanism to prevent loss of N via nitrate leaching (Munro, 1966). This phenomenon, termed biological nitrification inhibition (BNI), appears to be a relatively widespread phenomenon in tropical pasture grasses but also was found in several C3 plants, such as a perennial ryegrass that is used as a turfgrass (Moore and Waid, 1971; Subbarao et al., 2007, Subbarao et al., 2009). Reduction in nitrification could result in a decrease of N2O emissions from the process (Subbarao et al., 2009).

Turfgrass management practices can alter ecosystem C balance by affecting CH4 uptake and soil C sequestration. The few existing studies of CH4 fluxes suggest that the capacity of CH4 uptake is reduced in managed turfgrass systems when compared with native grasslands or forests (Kaye et al., 2004; Groffman and Pouyat, 2009). Lawns are either weak sinks or weak sources of CH4 (Groffman and Pouyat, 2009). Turfgrass management practices tend to enhance SOC storage by increasing above- and belowground production (Falk, 1976; Falk, 1980, Golubiewski 2006). Qian and Follett (2002) reported that golf fairways could sequester SOC at a rate of 1.0 Mg ha−1 yr−1 during the first 25 yr conversion from crop land or native grassland. Lawns within Denver (>25 yr of age) were reported to have almost twofold higher SOC densities than in nearby shortgrass steppe soils (Pouyat et al., 2009).

The difficulty of assessing GWPs by conducting short-term field experiments is that there are many components in GWP calculation and large spatial and temporal variations. One feasible way to estimate trace gas emissions and GWPs for ecosystems is computer modeling. DAYCENT is an ecosystem model that has the ability to simulate fluxes of N trace gases on a daily basis. The trace gas submodel of DAYCENT has been validated using data from various ecosystems and locations in the world (Del Grosso et al., 2002; Stehfest and Müller, 2004; Del Grosso et al., 2005; Pepper et al., 2005; Li et al., 2006; Adler et al., 2007; Del Grosso et al., 2008).

Previously, the CENTURY ecosystem model, which is the monthly version of the DAYCENT model, has been used to simulate SOC in golf course and home lawn conditions (Bandaranayake et al., 2003; Qian et al., 2003). The SOC submodel has been tested using soil C sequestration data from golf courses (Bandaranayake et al., 2003). We have also parameterized and applied the DAYCENT model to turfgrass systems to develop best management practices (BMPs) for maintaining high- and medium-quality Kentucky bluegrass lawns in Colorado (Zhang et al., 2013).

Although the DAYCENT model has been adjusted to simulate turfgrass ecosystems and validated using data of biomass, evapotranspiration, leaching, and soil temperature (Zhang et al., 2013), no research is available to compare the DAYCENT-predicted N2O flux with measured data in turfgrass lawns. The objectives of this research were (i) to validate the DAYCENT model’s ability to predict N2O emissions from turfgrasses by coupling field measurements of two published experiments with the DAYCENT simulations; (ii) to simulate the impact of different management practices, including a conventional management practice and a DAYCENT model-generated BMP, on N2O emissions; and (iii) to determine the net GWP of turfgrass management over time by combining energy expenses associated with turfgrass maintenance and N2O emissions simulated in current study with SOC sequestration estimated in previous study by Zhang et al. (2013).

Materials and Methods

DAYCENT Model Description

The DAYCENT model was developed based on the CENTURY model, which has been widely used in simulations of medium- to long-term (10 to >100 yr) changes in soil organic matter (SOM), plant productivity, and other ecosystem parameters for the major ecosystems in the world (Parton et al., 1987; Parton et al., 1993; Parton et al., 1994). The DAYCENT model uses daily time scale in modeling decomposition, nutrient flows, soil water, and soil temperature and has increased spatial resolution for soil layers. The main inputs of the DAYCENT model are (i) soil texture, (ii) daily weather data (maximum/minimum air temperature and precipitation), (iii) plant type, and (iv) management practices (e.g., irrigation and amount and timing of fertilizer applied).

Nitrous oxide emissions from nitrification and denitrification are modeled. Modeled N2O fluxes from nitrification are a function of soil NH4+ concentration, water-filled pore space (WFPS), temperature, and soil texture. Nitrous oxide emissions from denitrification are a function of soil NO3 concentration, WFPS, heterotrophic respiration, and soil texture. In the version of the DAYCENT model (DAYCENT 4.5) used in this study, denitrification is assumed to occur only when WFPS is above ∼55%, and the rate increases exponentially when WFPS increases from ∼55 to ∼90%. In the DAYCENT model, NH4+ is assumed to be distributed only in 0- to 15-cm soil because of its immobility, whereas NO3 is distributed throughout the soil profile. The SOC simulated by the DAYCENT model is within 0 to 20 cm of soil profile. The labile C availability is approximated by simulated heterotrophic respiration.

Simulation of the Field Experiments

In this study, measured N2O fluxes from experiments conducted by Kaye et al. (2004) and Bremer (2006) were used to evaluate the performance of the trace gas submodel. For each of these studies, we simulated turfgrass management practices, including irrigation, fertilization, and mowing. Detailed information on the DAYCENT model parameterization has been documented previously (Zhang et al., 2013). Previous land use before each experiment was simulated according to land use history.

In the experiment by Kaye et al. (2004), N2O fluxes were measured for 1-yr period on three turfgrass sites (one institutional lawn and two home lawns) dominated by Kentucky bluegrass (Poa pratensis L.) in Fort Collins, Colorado. Fluxes of N2O from soil were estimated by using static soil chambers (Mosier et al., 1991; Mosier et al., 1997). Sampling dates were approximately twice per month during the growing season (Apr.–Oct.) and monthly during the winter. Additional samples were collected before and after fertilization and irrigation events when fluxes are expected to be greatest. Fluxes were measured between 0900 and 1300 h MST, which is used to represent the average flux value for the day. The annual emission rates are calculated using linear interpolation between measurement dates (Kaye et al., 2004). The soil analysis results of the sites are shown in Table 1. Soil temperature and soil water content were measured on gas-sampling dates. Fertilizer applied in June and October totaled 110 kg N ha−1 yr−1. The institutional lawn was fertilized with urea (46–0–0). The two home lawns used commercial fertilizer (25–5–5) (Jirdon Agri Chemicals, Inc.). The N form in this commercial fertilizer is NH4+. Mowing was scheduled weekly, and clippings were returned to the site.

View Full Table | Close Full ViewTable 1.

Soil properties of the simulated experimental sites.

Site Sand Clay Silt Bulk density pH
% g cm−3
Kaye et al. (2004) experiment
 Kentucky bluegrass lawn A 47 30 24 1.15 7.5
 Kentucky bluegrass lawn B 74 12 14 1.21 7.7
 Institutional lawn 53 22 26 1.21 7.8
Bremer (2006) experiment
 Ryegrass lawn 32 24 44 1.20 7.2
Zhang et al. (2013) experiment
 Kentucky bluegrass lawn at Northern Colorado Water Conservancy District 54 29 17 1.27 7.2

The irrigation schedule was not documented in this experiment. The total amount of irrigation applied during the growing season is 54 ± 4 cm yr−1. To simulate the irrigation of the three Kentucky bluegrass lawns, we divided the total amount of sprinkler irrigation from May to October into each month according to the monthly evapotranspiration predicted by the DAYCENT model to mimic the common irrigation management in this area. All three sites were converted to lawns about 60 to 100 yr ago. To estimate the soil property affected by long-term management, turfgrass maintenance practices were modeled for 80 yr. The daily weather data were obtained from the online database of Colorado Climate Center (Station number 53005, Fort Collins).

The experiment of Bremer (2006) was performed on a perennial ryegrass (Lolium perenne L.) turf in Manhattan, Kansas with three fertilization treatments: urea at two rates (250 and 50 g kg N ha−1 yr−1) and ammonium sulfate at 250 kg N ha−1 yr−1. The measurement method of N2O fluxes was similar to the Colorado experiment (Kaye et al., 2004). Samples were collected weekly, and more frequent measurements were taken after fertilizer applications. Soil properties are described in Table 1. Mowing was conducted once or twice weekly at 7.5 cm. Irrigation was applied three times weekly to keep turfgrass from drought stress. Soil moisture and temperature at 5 cm were measured daily. Ammonium and nitrate concentrations in the top 10-cm soil were measured four times during growing season.

The site has been established with perennial ryegrass since 1960. We simulated turfgrass management for typical lawn conditions for 45 yr (from 1960 to 2005, at which time the field experiment was performed). Urea and ammonium sulfate are NH4+–type fertilizers; the DAYCENT model simulated both of them as NH4+ was added to the soil. The applications of herbicide and fungicide were not simulated. Weather data were provided by Kansas State University Research and Extension (http://www.ksre.ksu.edu/wdl/; station ID: Manhattan).

To simulate the effect of BNI of perennial ryegrass (Moore and Waid, 1971; Subbarao et al., 2007), we decreased the nitrification coefficient (a multiplier on the nitrification rates) from 0.8 (default value) to 0.1 in 0.1 increments and compared the annual cumulative emissions with the measured values. For Kentucky bluegrass lawns, the nitrification coefficient remained the default value 0.8 because no BNI capacity has been reported for Kentucky bluegrass.

To evaluate the model simulation effectiveness, correlation analysis was performed by comparing measured with simulated N2O emissions for the Colorado and Kansas experiments. The Pearson product-moment correlation coefficient (r) was calculated.

Long-Term Predictions of the Impact of Different Management Practices on Nitrous Oxide Emissions for a Kentucky Bluegrass Lawn

After validation of the DAYCENT model, we conducted long-term simulations to predict the impact of turfgrass management (fertilization and irrigation) on N2O emissions on a Kentucky bluegrass study site managed as a home lawn at the Northern Colorado Water Conservancy District (NCWCD). We selected this site for the long-term simulations because the site has previously been used for validating the DAYCENT model on clipping yield, leaf N content, evapotranspiration (ET), deep percolation, nitrate leaching, and soil temperature (Zhang et al., 2013). Long-term management effects on turf quality, nitrate leaching, and C sequestration have been determined, and best N application rates have been generated for this site. The soil on the site was a Fort Collins loam (Table 1).

To evaluate the impact of irrigation on N2O emissions, we used three different levels of potential evapotranspiration (PET) replacement (60, 80, and 100% PET) to predict the effect of irrigation under constant fertilization of 90 kg N ha−1 yr−1 for 50 yr. Irrigation was scheduled every 3 d. To predict the effects of fertilization, we compared two fertilization scenarios: a constant N rate of 150 kg N ha−1 yr−1 and the best N fertilization rates developed by Zhang et al. (2013) for the site (the DAYCENT model generated BMP); both of the scenarios are under 100% PET irrigation. The DAYCENT-generated BMP of N fertilization was 240 kg N ha−1 yr−1 1 to 3 yr after establishment, 140 kg N ha−1 yr−1 3 to 6 yr after establishment, and 110 kg N ha−1 yr−1 7 to 17 yr after establishment. The N fertilization rate was further reduced gradually until reaching 50 kg N ha−1 yr−1 at 40 to 50 yr after establishment. The fertilizers used in the simulation were NH4+ type, and all clippings were returned.

Global Warming Potential Calculations

Based on the results of the long-term simulations, we calculated GWP for the Kentucky bluegrass lawn at NCWCD for two management scenarios: (i) the DAYCENT model–predicted BMP N fertilization rate with 100% PET irrigation and (ii) a conventional fertilization rate of 150 kg N ha−1 yr−1 with 100% PET irrigation.

The output from the long-term simulations, including N2O emission rates and net C sequestration rates, were converted to CO2 equivalents and used for GWP estimation. Methane uptake was ignored in the GWP calculation because of the minimal uptake or emission of CH4 in our simulation results (data not shown). Energy costs from maintenance (mowing, irrigation, and fertilization) were estimated by using published data as described below.

We assume a typical motorized walk-behind mower was used with gasoline consumption of 0.00094 L m−2 (Sahu, 2008). The emission of gasoline combustion is 2347.4 g CO2 L−1 (USEPA, 2011). Mowing was conducted weekly in this area from April to October (28 times yr−1). The annual GWP from mowing was calculated to be 61.6 g CO2 m−2 yr−1.

The irrigation water used in lawns is mainly from city potable water supply system in the city of Fort Collins, Colorado. The energy cost of Fort Collins’ water supply was 0.125 kWh m−3, including water treatment and distribution (Tellinghuisen, 2009). The CO2 emissions coefficient for electric utilities for Colorado is 929 g CO2 kWh−1 (U.S. EIA, 2001). The annual total emission from treating and distributing irrigation water is estimated at 116.0 g CO2 m−3. The total amount of irrigation water was predicted by the DAYCENT model.

Estimates of GHG emissions from manufacture and transportation of fertilizer are 3.3 to 6.6 g CO2 g−1 N, 0.37 to 1.1 g CO2 g−1 P, and 0.37 to 0.73 g CO2 g−1 K, respectively (Lal, 2004). We use means of 4.8, 0.73, and 0.55 g CO2 g−1 for N, P, and K, respectively, in our calculation. The application rates of P and K are based on the N application rate and the element ratio in the fertilizer. A common lawn fertilizer by weight is 29% N, 3% P, and 4% K (Zirkle et al., 2011). Regarding pesticide application, we used the averages of annual GWP of pesticide on turfgrass (11.7 g CO2 m−2 yr−1), which are estimated by Zirkle et al. (2011).

Results and Discussion

Validation of Simulated Nitrous Oxide Emission from Kentucky Bluegrass Lawns

The predicted annual cumulative N2O emissions from Kentucky bluegrass lawns were within 16% of the observed values for the study conducted by Kaye et al. (2004) (Fig. 1). The observed annual rate of N2O emissions of the institutional lawn was approximately 42% higher than the other two lawns, probably due to its higher SOM content (Christensen and Christensen, 1991; Merino et al., 2004; Li et al., 2005). The simulated trends of daily fluxes are acceptable; Pearson’s r equaled 0.57, 0.60, and 0.53 for Home lawn A, Home lawn B, and Institutional lawn on daily basis, respectively (Fig. 2). The observed and simulated results showed that N2O fluxes increased dramatically right after fertilization, although the DAYCENT model underestimated these peaks. The observed high fluxes in February 2001 were not simulated by the DAYCENT model, which likely resulted from the failure of simulating high soil water content at the time of soil thawing.

Fig. 1.
Fig. 1.

The comparison of measured and simulated annual N2O emissions from three lawns in Colorado. Measured data were from the experiment conducted by Kaye et al. (2004).

Fig. 2.
Fig. 2.

The comparison of measured and simulated daily N2O fluxes from three lawns in Colorado. Measured data were from the experiment conducted by Kaye et al. (2004). Arrows indicate the fertilization dates.


Although the simulation of daily N2O fluxes needs improvement, the DAYCENT model was able to predict annual cumulative emissions from Kentucky bluegrass lawns in Northern Colorado. The DAYCENT model is an intermediate complexity biogeochemical model that only requires inputs that are relatively easy to obtain. As described by Del Grosso et al. (2008), the accuracy of the DAYCENT model in simulating daily fluxes of N2O might be not very high compared with that of more complex mechanistic models that require much more detailed inputs; however, it is more practical to use the DAYCENT model to assess cumulative N2O emissions.

Our DAYCENT model simulation results suggested that nitrification was the main source of N2O emissions for the three lawns (>93% of total N2O). The proportion of N2O from nitrification and denitrification is a function of O2 availability that is affected by soil water status. Nitrification is active when soil water content is relatively low, and denitrification becomes the main source when soil is under anaerobic conditions. With frequent and relatively light irrigation, the three Kentucky bluegrass lawns (medium-textured soils) emitted most of N2O through nitrification process because the soils were usually under aerobic conditions in the semiarid area.

Validation of Simulated Nitrous Oxide Emission from Ryegrass Lawns

In the perennial ryegrass experiment, the observed annual N2O emissions showed little difference between treatments of urea and ammonium sulfate at rate of 250 kg N ha−1 yr−1. Approximately 50% more N2O emissions were found in high-N treatments (250 kg N ha−1 yr−1) than in low-N treatments (50 kg N ha−1 yr−1). Initially, the DAYCENT model overestimated annual emission rates by 218, 210, and 189% for treatments of ammonium sulfate at high rate, urea at high rate, and urea at low rate, respectively. Because perennial ryegrass has pronounced and persistent effects of BNI (Moore and Waid, 1971), we modified the nitrification rates to the effect of BNI by setting the nitrification coefficient in the model to 0.3. As a result, the simulated annual emissions for three N treatments were approximated within 8% of the observed values (Fig. 3). The Pearson’s r of predicted vs. measured daily fluxes of N2O was 0.50, 0.65, and 0.78 for treatments of ammonium sulfate at high rate, urea at high rate, and urea at low rate, respectively (Fig. 4a and b). Our simulation indicated that soil nitrification was reduced by 50 to 64% after the modification of nitrification coefficient. The reduction of total N2O emissions by BNI effects was similar to that found in a tropical grass Panicum maximum pasture (∼50%), which was detected at the same level of BNI capacity as ryegrass (Subbarao et al., 2007; Subbarao et al., 2009). Although nitrification was suppressed, the DAYCENT-simulated N2O emitted from nitrification was still the major component of the total N2O emissions from soil (66% for the high-N treatment and 80% for the low-N treatment).

Fig. 3.
Fig. 3.

The comparison of measured and simulated annual N2O emissions from a perennial ryegrass turf with three N treatments (AS, ammonium sulfate, 250 kg N ha−1 yr−1; UH, urea, 250 kg N ha−1 yr−1; UL, urea, 50 kg N ha−1 yr−1). Measured data were from an experiment in Kansas (Bremer, 2006).

Fig. 4.
Fig. 4.

The comparison of measured and simulated daily N2O fluxes. Perennial ryegrass was fertilized at N rates of 250 kg N ha−1 yr−1 (a) or at 50 kg N ha−1 yr−1 (b) using ammonium sulfate or urea. Measured data were from the experiment conducted by Bremer (2006). Arrows indicate the fertilization dates.


Soil temperature was correctly simulated in the growing season (Fig. 5a and b). The underestimation of soil temperature in winter (up to 7.8°C) likely results from the underestimation of the insulating effect of snow cover. Soil water content, directly related to WFPS, was not well simulated in the winter by the DAYCENT model (Fig. 5c), likely due to difficulty in modeling freezing and thawing events, which resulted in the mistiming of N2O peaks and in inaccurate estimation at the peaks. Simulated soil ammonium and nitrate concentrations for four measurement days in 2004 were acceptable, but the concentrations in the high N rate treatments were overestimated by the model (data not shown).

Fig. 5.
Fig. 5.

The comparison of measured and simulated daily maximum (a), minimum soil temperature (b), and soil volumetric water content (c). Soil temperature and water content was measured at 5 cm depth. Simulated results were the model output of the 2 to 5 cm soil layer.


Negative N2O flux or N2O uptake, which was observed in this ryegrass study, has been found in various other ecosystems (Chapuis-Lardy et al., 2007). However, it is still in debate whether negative values should be treated as errors or as measurement “noise” (Chapuis-Lardy et al., 2007). Because the mechanism of soil uptake of N2O is unclear, the DAYCENT model does not simulate the uptake of N2O.

Long-Term Effects of Different Irrigation Levels on Nitrous Oxide Emissions

Soil water status and soil N dynamics are closely related to irrigation. The long-term predictions of N2O emissions for a Kentucky bluegrass lawn irrigated with 60% PET, 80% PET, and 100% PET replacement are shown in Fig. 6a. Our simulation results indicated that 60% PET irrigation resulted in the highest annual N2O emissions over 30 yr, which was approximately twice as high as that of 100% PET irrigation applied with the same amount of fertilizer in most of the years. In contrast, in the last 10 yr of our simulation, annual emissions were slightly higher in 100% PET irrigation. The reason was likely that 60% PET replacement resulted in retarded growth of turfgrass and a reduction in N uptake, allowing more mineral N to accumulate in the top soil layer (Fig. 6b). When there was abundant mineral N in the soil for old turfgrass stands (>40 yr in Fig. 6b), annual N2O emission rates were not very different for the three irrigation levels. This result might be explained by simulated soil WFPS. Water-filled pore space was closely related to N2O emissions (Smith et al., 1998; Dobbie et al., 1999). Daily average WFPS in our simulation of 100% PET irrigation ranged from 0.4 to 0.5 (WFPS of 0.5 corresponds to the field capacity) for most of the growing season. Water-filled pore space for 60% PET irrigation mainly fluctuated between 0.25 and 0.5. According to Parton et al. (2001), in the DAYCENT model, the effect of WFPS on nitrification is highest when WFPS is approximately 0.4 in medium-textured soil. The effect drops from 1.0 to 0.8 when WFPS decreases from 0.4 to 0.3. Changing irrigation from 60 to 100% PET probably results in little difference on nitrification rate in the sandy clay loam soil when NH4+ is not a limiting factor. The majority of N2O emissions were modeled to be emitted through nitrification because denitrification was not simulated when WFPS was <0.55 (Parton et al., 2001). The highest annual N2O emissions in our simulation were predicted to reach 6.1 kg N ha−1 yr−1 at Year 46 in the 100% PET scenario, which is comparable to the estimated annual rate of 6.4 kg N ha−1 yr−1 in golf course fairways in Arizona with high N input in preceding years (Guilbault and Matthias, 1998).

Fig. 6.
Fig. 6.

Model predicted annual N2O emissions (a) and annual average soil mineral N content (b) in the top 10 cm soil for three irrigation levels for a Kentucky bluegrass lawn in Colorado. PET, potential evapotranspiration.


Long-Term Effects of Different Nitrogen Fertilization Regimes on Nitrous Oxide Emissions

Using a constant rate of N fertilization for many years is commonly found in the management of lawns. However, constant N rates may result in over- or under-fertilization and may cause environmental threats (e.g., nitrate leaching). In our long-term simulation, with fertilization of 150 kg N ha−1 yr−1 constantly, N2O emissions increased dramatically in the first 15 yr and leveled off at an average rate of approximately 5 kg N ha−1 yr−1 after 15 yr (Fig. 7). By applying the model-generated best N rates, annual N2O emissions were maintained in the range of 0.6 to 3.1 kg N ha−1 yr−1 from Year 10 to Year 50, which is approximately half the emission rate of using a conventional constant N rate. The N2O emissions were predicted to be higher in the first 10 yr using the model-generated best N rates because a larger amount of N was applied to help turfgrass to establish and exhibit high quality. In 50 yr, gradually reducing fertilization as the lawn ages would significantly reduce long-term N2O emissions by approximately 40% when compared with applying N at the constant rate (150 kg N ha−1 yr−1) without significant reductions in production.

Fig. 7.
Fig. 7.

Model predicted annual N2O emissions for two management scenarios (model predicted best N rate; constant N rate fertilizer at 150 kg N ha−1 yr−1).


Long-Term Effects of Different Management on Soil Carbon Sequestration

The long-term impacts of different management practices (irrigation and fertilization) on C sequestration of lawns have been assessed via model simulations in our previous publications (Zhang et al., 2013; Qian et al., 2003). The soil C sequestration in the surface 0 to 20 cm soil was estimated to be 32.1 to 32.6 Mg ha−1 over 40 yr (Zhang et al., 2013), and the amount of C sequestered was higher in the newly established lawn and declined as the turfgrass stand aged. The soil C sequestration outputs from the long-term DAYCENT model simulations were converted to CO2 equivalents and used for GWP estimation as described below.

Global Warming Potential Calculations

Total energy cost from turfgrass maintenance (the sum of GWPs for mowing, fertilization, irrigation, and pesticide application) accounts for the largest proportion of the total emissions (65–80%) in the first decade for the BMP and conventional management scenarios. Mowing, irrigation, and fertilization individually contribute to nearly equal proportions of emissions in first decades in the scenario of conventional management (Fig. 8a). With constant 150 kg N ha−1 yr−1 fertilization, the GWP of N2O emissions increases to around 2250 g CO2 m−2 decade−1 (half of the total positive GWP). In the scenario of BMP predicted by the DAYCENT model, the total emissions tend to decrease over decades (Fig. 8b). Because fertilization rate is reduced through time, the energy cost of fertilization decreases, whereas the GWP of N2O emissions is kept stable at an average rate of 1210 g CO2 m−2 decade−1.

Fig. 8.
Fig. 8.

Estimated global warming potentials (GWPs) for lawns using conventional management (a) and best management practices (b) generated by the DAYCENT model for a lawn to maintain high quality.


The GWPs of C sequestration vary slightly for the scenarios of constant N rate and model-predicted BMPs for a high-quality lawn. It is predicted that nearly 4000 g CO2 m−2 C could be sequestrated in the first decade. The sequestration rates gradually decrease as the lawn ages. The soil used in our simulation is a sandy clay loam with 29% clay. Soils with higher clay content could probably sequestrate substantially more C over the long term because clay particles provide greater protection to SOM (Bandaranayake et al., 2003).

The model predicted that the amount of water needed for irrigation ranged from approximately 40 to 80 cm per year, which resulted in GWPs of 460 to 930 g CO2 m−2 decade−1. Because Fort Collins Water Utility relies on gravity-fed surface water supplies, the GWP of irrigation is likely to be half or less than half of those of some cities in the South Metro area of Colorado that use substantial amounts of energy for pumping groundwater from the Denver Basin aquifers (Tellinghuisen, 2009). Methane uptake was not included in our calculation. Lawns in Colorado were found to be sinks of CH4, with annual uptake rate of 0.15 g C m−2; this rate is equal to 42 g CO2 m−2 decade−1, which is negligible compared with N2O emissions (Kaye et al., 2004). In Maryland, Groffman and Pouyat (2009) observed that lawns are either weak sinks or weak sources of CH4.

By considering aspects such as soil C sequestration, turfgrass maintenance energy costs, and N2O emissions, the Kentucky bluegrass lawn could change from a sink to a weak source of GHG emissions by about 20 to 30 yr after establishment (Fig. 8). One study conducted in California has shown that turfgrasses serve as either sources or sinks of global warming depending on fertilization rates (Townsend-Small and Czimczik, 2010). However, the fertilization rate of 750 kg N ha−1 yr−1 in that experiment is considered extremely high and is rarely used in the turfgrass industry (Law et al., 2004).


Computer modeling provides a relatively easy way for assessing the GHG budget for ecosystems. Our study showed that the DAYCENT model can properly simulate the annual N2O emissions for Kentucky bluegrass and perennial ryegrass lawns. The simulation for perennial ryegrass indicates that BNI might play an important role in reducing N2O emissions. During the first 30 yr, a Kentucky bluegrass lawn was predicted to emit more N2O when irrigation was applied by replacing 60% PET than that of 100% PET. The model suggests that gradually reducing fertilization as the lawn ages from 0 to 50 yr would significantly reduce long-term N2O emissions. In the conventional and the model-generated BMP scenarios, GHG emissions from maintenance account for 50 to 80% of the total emissions in each decade (Fig. 8). Keeping a constant high fertilization rate of 150 kg N ha−1 yr−1 could substantially increase GWP of N2O to half of the total emissions. Our DAYCENT model generated best N rates for a high-quality lawn could help reduce the amounts of the positive GWP by 25% in 50 yr compared with those of the conventional N rates but maintain a similar soil C sequestration rate.


This research is partially supported by US Golf Association. The authors thank Dr. W.J. Parton and Dr. A.J. Koski for critical discussion and Ms. Sarah Wilhelm and Ms. Cindy Keough for technical support.




  • Assigned to Associate Editor Stephen Del Grosso.

Be the first to comment.

Please log in to post a comment.
*Society members, certified professionals, and authors are permitted to comment.