Int J Biometeorol (2017) 61:103–113 DOI 10.1007/s00484-016-1194-z

ORIGINAL PAPER

Carbon dioxide fluxes from Tifway bermudagrass: early results David L. Cotten 1,2 & G. Zhang 1

&

M. Y. Leclerc 1 & P. Raymer 3 & C. J. Steketee 3

Received: 19 December 2014 / Revised: 28 May 2016 / Accepted: 29 May 2016 / Published online: 15 June 2016 # ISB 2016

Abstract This paper reports for the first time preliminary data on carbon uptake of warm-season turfgrass at a well-managed sod farm in south central Georgia. It examines the changes in carbon uptake from one of the most widely used warm-season turfgrass cultivars in the world, Tifway Bermudagrass. It elucidates the role of canopy density and light avalaibility on the net carbon uptake using the eddy-covariance technique. Preliminary evidence suggests that turfgrass is effective at sequestering carbon dioxide during the summer months even when the canopy is being reestablished following a grass harvest.

Keywords Turfgrass . Carbon flux . Eddy covariance . Sod farm

* M. Y. Leclerc [email protected]

Introduction Background Climate extremes along with their impact on ecosystem’s ability to capture carbon are steadily rising. The role of turf in both the carbon and hydrological cycles has remained largely ignored despite the fact that turf ecosystems are ubiquitous in the form of home lawns, golf courses, and soccer fields, covering an estimated 160,000 km2 in the continental USA alone (Bartlett and James 2011; Milesi et al. 2005). This information gap is especially relevant for warm-season grasses (Fissore et al. 2012; Wu and Bauer 2012) which grow over a long season, and for which there has been a paucity of studies in the field. Addressing the ability of warm-season grasses to sequester carbon is thus relevant to both the national biogeochemical inventories and to the turfgrass industry. Owing to their long growing season (approximately 8 months a year in the southern USA), warm-season varieties of turfgrass are of particular significance because of their relatively large potential to sequester carbon due to their dense root structure and canopies. The present study fills a gap amidst the modest corpus of studies on the subject as it focuses its attention on warm-season grasses (Bremer and Ham 2005; Wu and Bauer 2012). The few studies on turfgrass cultivars were performed using a variety of techniques and had wideranging results. Very few studies used the eddy- covariance technique and instead focused on an assortment of methods ranging from vegetation sampling to remote sensing.

1

Laboratory for Environmental Physics, The University of Georgia, 1109 Experiment Street, Griffin, GA 30223-1797, USA

2

Center for Geospatial Research, Geography, The University of Georgia, 210 Field Street, Athens, GA 30602-5026, USA

Previous studies

3

Institute of Plant Breeding, Genetics and Genomics, The University of Georgia, 1109 Experiment Street, Griffin, GA 30223-1797, USA

The majority of turfgrass carbon studies were performed on cool season grasses and therefore are expected to

104

have lower carbon uptake than warm season grasses due to their physiological differences. One cool-season grass based carbon study by Fissore et al. (2012) was built on vegetation sampling coupled with allometric and biogeochemical models to estimate fluxes. They found the mean of 360 single-family household lawns in Minneapolis-Saint Paul, MN, sequestered on average 0.51 kg of C m−2 year−1. However, since the study area included trees, the carbon fluxes measured in this study were contaminated by their presence. Wu and Bauer (2012), using remote sensing techniques in conjunction with in situ measurements of Kentucky bluegrass, found that the average net production of golf course grass was much higher than that of lawn grasses, 1.1 kg of C m−2 year−1 relative to 0.771 kg of C m−2 year−1. This difference could well be attributed to different management techniques used by homeowners versus the more rigorous routines used by golf course superintendents. Allaire et al. (2008) used closed chambers to measure average CO2 fluxes over Kentucky bluegrass in urban areas every 7–14 days. Their results were nearly twice as large as those found by Wu and Bauer (2012) using the same cultivar, 0.058 mg of C m−2 s−1 which corresponds to roughly 1.8 kg of C m−2 year−1 (Ham et al. 1995; Hull 2000; Mariko 2007). A study done in California using soil cores to measure organic carbon was performed by Townsend-Small and Czimczik (2010). Their results found ornamental lawns in California sequestered 0.14 kg of C m−2 year−1, which was close to that observed in regrowing forests found in the northeastern USA. Another study based only on below ground measurements with the caveat that the results emphasized the high-end of the carbon uptake owing to optimum growing conditions showed that warm season turfgrass ecosystem s can sequester up to 0.1 kg of C m−2 year−1 (Selhorst and Lal 2011). Past studies focusing on warm season grasses typically involved bermudagrass, although warm season grasses were far less studied than their cool season counterparts. Dugas et al. (1999) measured carbon dioxide fluxes from bermudagrass using a Bowen ratio system, which is of particular relevance to the present study. The Bowen ratio method is similar to the eddy covariance technique but is not as robust due to its slower data collection time, one data point every every 2 seconds versus 10 data points every second. They found that bermudagrass released 0.1 kg of C m−2 year−1 from February to November 1993, where the turfgrass was a net source of CO 2. However, the bermudagrass sequestered 0.76 kg of C m−2 year−1 the following year, 1994, where measurements were taken from March to November. The authors attributed the shift from a source to a sink to the turfgrass being planted in 1993.

Int J Biometeorol (2017) 61:103–113

Rationale The goal of the present paper is to quantify the carbon sequestration potential of a warm season turfgrass cultivar with quasi-ideal micrometeorological fetch conditions present at a turf farm. The method used to obtain the data, i.e., the eddy covariance method (hereafter EC), is a proven and advanced technique used to obtain accurate aboveground net exchange of carbon dioxide and water vapor. This technique is considered advanced because it requires strict quality control methods and certain environmental conditions to obtain accurate results. It is an aerodynamic method predicated on determining the CO2 balance between CO2-rich downdrafts and CO2-poor updrafts. In addition, it does not disturb the structure of the turfgrass and provides continuous carbon and water fluxes. For more information on this powerful method, the reader is referred to papers, and the modus operandi are abundantly described in papers by Leclerc and Thurtell 1990; Vesala et al. 2007; Rannik et al. 2011; Aubinet et al. 2012; Leclerc and Foken 2014. By using the eddy-covariance method, the results of this manuscript are more robust than some of the aforementioned techniques because of its nearly continuous monitoring and ability to take measurements without disturbing the ecosystem. Additionally, the eddy-covariance method allows for the removal of contaminated data through the use of quality assurance/quality flags (Vickers and Mahrt 1997). Therefore, all results presented are accurate representations of this cultivar because the effects of all outside sources were removed.

Materials and methods Field and instrumentation The field experiment was performed at the Middle Georgia Super-Sod farm in Fort Valley, GA. The field chosen as the research site was large (∼1 km2) and flat, and contained only one variety of turfgrass, Tifway bermudagrass (Fig. 1a), meeting the quasi-ideal (i.e., large, flat, homogeneous) conditions required for EC measurements. Data collection began in July, during regrowth after a previous harvest of turf, when the canopy coverage was approximately 60 % and continued while the canopy coverage was 100 %. The sod was reharvested a week following the experiment; therefore, the area measured was completely covered, >95 %, for the entire last month (September) of the present experiment. Two eddy - covariance systems were deployed in the field with instrumentation located at 1.5 m above ground level (AGL) on tripods and facing nearly opposite directions (Fig.

Int J Biometeorol (2017) 61:103–113

105

a 0 East Tower Points to 240

320 200

West Tower Points to

80

b

covariance system, all at 10 Hz. The data logger also recorded diagnostic parameters of the CSAT3 and Li-7500 for analysis of data quality. Belowground measurements were also made at the West Tower site which included soil temperature, volumetric soil water content, and heat flux. Soil temperature and water content were measured at depths of 0.02, 0.05, 0.1, and 0.2 m. Soil temperature data was acquired using four custom-built chromel-constantan thermocouples and soil volumetric water content measured using time-domain reflectometry sensors (CS615, Campbell Scientific, Logan, UT). Soil heat flux was measured using two soil heat flux plates (Campbell Scientific HFT-3.1) at 8 cm depth, 0.9 m apart. Meteorological measurements were made using an automatic weather station starting on August 22, 2013. Measurements included rainfall, incoming solar radiation, wind speed and direction, temperature, and relative humidity. Only soil and eddy - covariance data were available on July 1 at the turf site, with radiation and rainfall becoming available by August 22, 2013 (DOY 234). To fill in the missing radiation data a nearby, ∼10 km, Georgia Weather Net (GWN) site located in Fort Valley, GA, was used to provide on-site radiation values. The 30min averages of incoming solar radiation measured in W m−2 at the turf site versus the GWN site for the end of August throughout September demonstrated similar results, R2 value = 0.95. Therefore, the radiation data for the GWN site was used in the calculations throughout the manuscript. Data processing

Fig. 1 a Two-tower setup used at the Super Sod Farm in Fort Valley, GA, with a 100 m footprint represented for each tower, Image credit: Google Earth 2014; b A close-up view of one of the eddy covariance systems, East Tower

1a), one located on the west side (West Tower) and another located on the east side (East Tower) (Fig. 1b). The height and locations of each tower were chosen to minimize any interference from sources outside of the turf field, e.g., all roads and other vegetation were at least 150 m away. Each system consisted of a three-dimensional sonic anemometer (model CSAT3, Campbell Sci., Logan, UT, USA) and a fastresponse open-path CO2/H2O gas analyzer (model Li-7500, Li-Cor Inc., Lincoln, NE, USA). A data logger (model CR1000, Campbell Sci., Logan, UT, USA) was used to record the three-dimensional velocity components, virtual sonic temperature, and CO2 and H2O concentrations in each eddy-

The data from each tower were divided into 30-min runs, and the quality of each run was assessed with the aid of the CSAT3/Li7500 diagnostic parameters. Data with poor quality indicated by the diagnostic parameters, usually due to rain or the accumulation of dew or fungus on the sensors’ optical/sonic paths, were removed from the analysis. Spikes in the data were detected and removed using the method proposed by Vickers and Mahrt (1997). The planar fit (PF) method (Wilczak et al. 2001) was applied to correct for any sonic anemometer tilt errors that could have occurred during the installation of the equipment. Then, the data was linearly detrended (Rannik and Vesala 1999) to remove any temporal trends in each individual run. The resulting 30-min runs of data were used to calculate variances and covariances of CO2, H2O, temperature, and wind. From these statistical quantities, fluxes of carbon dioxide and water vapor were calculated. Density corrections were applied to the calculations of the CO2 and latent energy fluxes using the technique described in Webb et al. (1980). In order to avoid contaminated data due to flow distortion by the eddy covariance towers and instruments themselves,

106

the angle of attack of the 30-min mean wind vector on the sonic transducers was verified for each run. Using this mean vector data, wind directions between 80° and 320° were discarded for the West Tower to eliminate any contamination effects caused by the insufficient fetch. Specifically, the pecan forest 150 m to the south, the storage area to the east, and the ditch located to the west of the tower. For the East Tower, data with wind in the range of 0° and 200° was also discarded to avoid any distortions from the tower and any contamination from the abandoned lots located directly east of the farm. Figure 1a shows the positions of the two towers with their respective fetches. The arcs have a radius of ∼100 m to show the most common areas Bseen^ by the towers, as well as demonstrating why other directions were excluded. The towers only overlapped in wind direction between 320° and 360°, which was the least common wind direction, accounting for only 4 % of all the wind components. After removing unusable data due to poor weather conditions and using the diagnostic parameters as a guide, the data from the two towers only overlapped for 35 of the 4416 half-hour flux windows for this study. When the two datasets were compared at each 30-min run where they overlapped, they agreed within ∼55 % of each other. This value can be explained because even though the large-scale structure of the field was homogenous, some smaller scale areas can differ greatly from one another. Furthermore, since the footprint of these towers, the effective upwind source area sensed by the observation, can vary from a few meters to around ∼150 m during daytime measurements, direct comparisons of the two towers should be performed with caution (Kljun et al. 2004; Leclerc and Thurtell 1990; Leclerc and Foken 2014). Yet, these comparison errors as well as the standard deviations were calculated for each 30-min block at the same time, e.g., all 13:30 times for 1 month are averaged and have their standard deviations calculated. Even though footprints of the towers rarely reached beyond 100 m except for the some nights with very stable atmospheric conditions, the area between 80° and 200° in the dataset combined from both west and east towers were excluded to avoid any outside interference either due to carbon sources or obstructions effecting turbulent flow, which only resulted in the loss of 35 % of all the wind components. Overall, only 31 % of the 4416 possible half-hour flux measurements were used for data analysis; the rest were excluded to ensure no outside carbon sources nor incorrect equipment readings could interfere with the measurements. More data was lost on nighttime (18:00 LST to 6:30LST) conditions, 30 % more than daytime, due to condensation on the IRGA sensor and low wind conditions. Gaps in the missing data were not replaced by proxy data as is sometimes done in micrometeorology. To ensure no data biases were caused by this difference in data, averages were calculated for each 30-min time stamp individually before determining the daily average for the months. To examine the relationship between incoming solar radiation and aboveground CO2 fluxes, the data from the two

Int J Biometeorol (2017) 61:103–113

systems were matched by their time-date stamp and sorted into three groups, July, August, and September. This data was combined with radiation data from the nearby (<10 km) Georgia Weather Net station and the radiation data that did not have any associated flux measurements were discarded for this particular examination. Only radiation data with values greater than 20 W m−2 were used to eliminate any issues that could arise from low light conditions. The relationship between half-hour CO2 fluxes, i.e., net ecosystem exchange (NEE) (μmol CO2 m−2 s−1), and photosynthetically active solar radiation (PAR) (μmol photons m−2 s−1) were modeled using the widely used Michaelis-Menten equation, Eq. 1 (Michaelis and Menten 1913; Ruimy et al. 1995; Frolking et al. 1998): NEE ¼

a⋅PAR⋅NEEsat þ Re a⋅PAR þ NEEsat

ð1Þ

where a is the apparent quantum yield or the initial slope of the light response curve (μmol CO2 μmol−1 photons), NEE sat (μmol CO 2 m −2 s −1 ) is the saturation value of NEE at an infinite light level, and Re is the ecosystem respiration in daytime conditions. Thus, utilizing the rectangular hyperbolic saturation curve derived from fitting the half-hourly data to the Michaelis-Menten equation yields the maximum NEE potential which allows for future comparisons with other ecosystems (Frolking et al. 1998). The curve was calculated using Microsoft Excel’s solver function, where Eq. 1, NEE, and PAR were used as the initial variables and the output were NEEsat, a, and Re. Canopy coverage To address the change in canopy coverage during the experiment, a light box was used to provide images of specific locations at different times during the study (Haselbauer et al. 2012). The use of a light box ensured the canopy was lit with the same intensity and distribution of light during each visit. This method was used to characterize canopy coverage, as the turf is too short for other conventional methods to measure canopy characteristics. The light box consisted of a ∼1-m3 metal box with four compact fluorescent light bulbs positioned at the inside top four corners of the box. A hole was located at the top center of the box to allow for a digital camera to be securely attached, facing directly downward toward the canopy. The same camera and settings, as well as the same locations in the field, were used during each visit to ensure consistency in the images. Four consistent locations, all chosen to be within the average upwind direction of each tower, were used to take digital images and then image data was averaged for each tower and standard deviations calculated. Images used in this evaluation were taken at 5, 13, 18, and 21 m from the flux system sensor position in the direction each sensor was pointed. The images were taken on August 1, 14,

Int J Biometeorol (2017) 61:103–113

and 22 and September 25 for both the East and West Tower locations, and additional images were taken on July 2 for the West Tower locations (Table 1). Estimates of percent green cover were obtained by analyzing digital images using Assess version 2.2, a software program from APS Press (American Phytopathological Society, St. Paul, MN) that specializes in disease image analysis and estimation of ground cover (http://www. apsnet.org). Prior to using the program, all images were carefully visually examined to ensure consistent quality and were cropped to eliminate unneeded pixels of the light box structure. Dividing the number of green pixels by the total number of pixels in the image gives the percent green cover value used in the paper. Percent cover was also visually evaluated by determining vegetative cover (regardless of tissue color). For example, if approximately 3/4 of the photo was covered by turfgrass, regardless if it was green, it would receive a 75 % rating. The percent green dropped from both locations between August 22 and September 25. This was due the turfgrass being cut earlier in the day, and our measurement being taken before the clippings were removed. This represents a prime example of why automated pixel analysis has some caveats and is still used in conjunction with visual inspection. The user was able to still measure canopy coverage through visual inspection and is why we used this metric as our guide for data analysis.

Results Soil heat flux and temperature Figure 2 displays the monthly average of diurnal variation of soil heat flux and standard deviations for the months of July, August, and September. The plots demonstrate how increasing the canopy coverage from 60 to nearly 100 %, see Table 1, as was the case from July to September can cause the soil heat flux underneath the canopy to decrease by a factor of two for both the heat loss at night and the heat gain during the day. These monthly averages show how the canopy acts as an insulator to the belowground from the aboveground components. Similar conclusions can be deduced from the monthly average soil temperatures. Figure 3a–c shows the monthly average of diurnal variation of soil temperature data at four depths: 2, 5, 10, and 20 cm. The figures are plotted using the same scales for sake of comparison of soil temperature among the different soil depths. As the canopy coverage increases and solar intensity starts to decrease, both the dynamic range of soil temperatures and their peaks decrease from July to September, where September shows the smallest difference between the 2 and 20 cm temperatures.

107

Net carbon uptake The two flux systems were positioned to ensure the acquired flux measurements pertained to the largest amount of upwind homogeneous Tifway bermudagrass area as possible, a requirement of the eddy flux method. Measurements from the two towers were combined to provide a more complete dataset. The daily averages of CO2 in milligrams per square meter stored or released by the cultivar can be determined from the 30-min fluxes (mg m−2 s−1) of carbon dioxide measured by flux systems on both towers and selected for analysis based on wind direction and data quality criteria. During the night, the turfgrass ecosystem is a source of CO2 due to the respiration from both the turfgrasses and the soil. The cultivar in question, during the daylight hours 6:30 LST to 18:00 LST, when the cultivar is photosynthetically active, has a negative flux of CO2. This negative flux equates to CO2 being removed from the atmosphere by the ecosystem. In nighttime conditions, the ecosystem shows a positive flux of CO2, i.e., a loss of carbon to the atmosphere due to respiration. As with the previous section, to observe how the change in canopy coverage impacts CO2 fluxes, the data were divided into three groups (July, August, and September) to further explore the relationship between canopy density and carbon uptake/release. These divisions were based on visual canopy cover derived from the light box imagery. Visual observations from July 1 to August 1 (DOY 181-213) showed that the canopy cover increased from 60 to 75 %, while for the majority of August (August 1–25, DOY 213–237), the canopy cover increased from 75 to 95 %. For the remainder of the experiment, from August 25 to September 30 (DOY 237-274), the canopy cover was greater than 95 %. In order to obtain the monthly average of CO2 flux diurnal variation for each month of the experiment, CO2 fluxes were averaged and standard deviations calculated at each half-hour window for the aforementioned monthly time frames. The bias due to the larger daytime dataset than the nighttime dataset was reduced since averages and deviations were taken at each 30min time stamp to determine the average flux at each time stamp for an entire month. These averages were then summed to yield an average daily value for the month. If daily averages were found for each day and then averaged for the month, the bias toward daytime values would be evident. For the month of July, the average number of half-hour fluxes available for each 30-min window was nearly identical for nighttime and daytime, being 4.6 and 4.3, respectively. The month of August yielded an average of 17.3 and 12.7 for nighttime and daytime, with September resulting in 7.2 and 5.0. For the month of July, 7.82 ± 0.47 g of CO2 m−2 per day were taken up by the turfgrass. In August and September, 11.81 ± 0.55 and 16.34 ± 0.61 g of CO2 m−2 per day were sequestered, respectively, showing similar increases in carbon dioxide uptake each month as the canopy coverage

108 Table 1 Averages and standard deviations of percent canopy coverage using two different methods calculated from four image positions located 5, 13, 18, and 21 m from each sensor location

Int J Biometeorol (2017) 61:103–113

Date

July 2, 2013 August 1, 2013 August 15, 2013 August 22, 2013 September 25, 2013

Percent green cover (Assess) East average

East std. dev.

N/A

N/A

56.7

15.8

West std. dev.

East average

East std. dev.

West Average

47.6

16.5

N/A

N/A

61

68.5

15.7

68

19

81

17 12

54.9

9.95

62.5

8.54

83

17

87

66.4

6.72

86.2

3.40

90

10

98

100

0

100

54.0

11.2

increased. The monthly averages of CO2 flux diurnal variation can be seen in Fig. 4a–c.

Influence of solar radiation on CO2 fluxes Figure 5a–c shows the relationship between NEE, PAR, and the Michaelis-Menten fit data with Pearson r values of 0.72, 0.70, and 0.72 for July, August, and September, respectively. These figures display the effect of radiation on CO2 fluxes from the sparse canopy for the month of July, the partial canopy for August, to the full canopy found during the month of September. As canopy coverage increases, so does the potential maximum amount of carbon that can be sequestered. Thus, not surprisingly, owing to a sparse canopy, the month of July had the smallest capacity for sequestering carbon yielding a maximum net ecosystem exchange potential of 1.24 (mg CO2 m−2 s−1), while August and September showed increases each month, with 1.56 and 1.85, respectively. Fig. 2 Monthly averages and standard deviations of the diurnal variations of soil heat flux for July, August, and September 2013

West average

Percent cover (Visual)

42.5

17.4

West std. dev. 6.3

2.5 0

Discussion This paper has provided an early insight on the behavior of a widely used warm season turfgrass ecosystem during its growing season. It has done so using a robust internationally used method of measuring carbon fluxes into and out of an ecosystem in question, the eddy flux method. Data analysis has followed rigorous data screening protocols removing unreliable data. Our site was specifically selected to eliminate any outside carbon sources, turbulent interference, and complex hilly or patchy terrain. Waiting for a larger corpus of data, this preliminary study suggests that, for the period of the study, a net amount of carbon was captured by the turf ecosysyem during the growing season. However, it should be kept in mind that the amount of carbon captured is tightly coupled to canopy coverage. When the canopy coverage was roughly 60 %, just 7.82 ± 0.47 g of CO2 m−2 day−1 was sequestered during July. This amount was more than doubled to 16.34 ± 0.61 g of CO2 m−2 day−1 during September once the canopy was fully covered.

Int J Biometeorol (2017) 61:103–113 Fig. 3 a July, b August, and c September 2013 monthly average soil temperatures diurnal variations at depths of 2, 5, 10, and 20 cm

109

110

Int J Biometeorol (2017) 61:103–113

Fig. 4 Monthly average diurnal variation of WPL-corrected carbon dioxide fluxes (mg m−2 s−1) for a July, b August, and c September 2013; error bars represent 1-sigma standard deviations and average error between the two towers

The results for July and September extrapolated over the entire year yield 0.8 and 1.62 kg of C m−2 year−1. These are comparable to the values from golf course maintained

Kentucky bluegrass at 1.1 kg of C m−2 year−1 but are larger than the results of homeowner managed Kentucky bluegrass at 0.771 kg of C m−2 year−1 (Wu and Bauer 2012).

Int J Biometeorol (2017) 61:103–113

111

Fig. 5 Photosynthetically active radiation as a function of carbon dioxide flux for a July, b August, and c September 2013, fit to a Michaelis-Menten model yielding Pearson r values of 0.72, 0.70, and 0.72 for July, August, and September, respectively

And the results yielded from the current study are less than Allaire et al. (2008) which was 1.8 kg of C m−2 year−1

measured from Kentucky bluegrass in an urban environment using chamber measurements.

112

Even more variation is presented when compared with previous bermudagrass measurements. Dugas et al. (1999) found that bermudagrass released 0.1 kg of C m−2 year−1 in the first year and sequestered 0.76 kg of C m−2 year−1 the following year. These results do not match very well with this current study because Dugas et al. (1999) claimed that the turfgrass was a source of carbon the first year because it was just planted, yet in the current study even when the canopy coverage was only ∼60 % the turfgrass showed a net uptake of CO2. The present study has examined the net amount of carbon stored in the plant-soil-root turf ecosystem and removed through mowing, during the most active growth period of the plant cycle. The present initial study of net ecosystem exchange of warm-season grass suggests that the maximum net ecosystem exchange to be intimately coupled to the canopy density where the canopy of 60 % coverage in July had the smallest potential for capturing carbon at 1.24 mg of CO2 m−2 s−1, and the canopy of almost full coverage in September showed a nearly 50 % increase to 1.85 mg of CO2 m−2 s−1. These results demonstrate the potential this warm-season turfgrass has to sequester carbon. Comparing the light use efficiency (LUE) found for bermudagrass in this study to previous studies was difficult due to lack of data in the literature. Zhang et al. (2012) found average NEEmax values of 0.9 mg of CO2 m−2 s−1 for croplands during the summer months and values of 0.34 for the steep regions during the same period. The results found in this study for bermudagrass were nearly twice as high compared to the crop land and six times as great compared to the steppe region; this is most possibly due to the abundance of water available in our turfgrass study. While this current study examined several months throughout the growing season, a more comprehensive, year-long experiment is needed to produce a true carbon balance so that the full range of temperature, day length and different canopy stages are included. The present study serves as a starting point to help understand and characterize commonly used turfgrass in the southeast. While the results of the study, done in quasi-ideal conditions of maintenance, fertilizer application and irrigation at a sod farm, these results are likely to represent the upper limit of carbon sequestration from household lawns as the maintenance regime is likely to vary from homeowner to homeowner. However, they are likely to represent broad agreement with other well maintained lawns such as recreational fields, golf courses, commercial properties during the peak of the growing season.

Conclusion The present study has presented a first preliminary overview of the net carbon uptake from one of the most widely used turf cultivars (Tifway Bermudagrass) in the southeast at a well

Int J Biometeorol (2017) 61:103–113

maintained sod farm and it has done so primarily during the growing season. It elucidates the role of canopy density and light avalaibility on the net carbon uptake using the eddycovariance technique. Preliminary evidence suggests that turfgrass is effective at sequestering carbon dioxide during the summer months even when the canopy is being reestablished following a grass harvest. Additional year-long and multi-year study should be done so as to produce a true carbon balance that can be incorporated in various models. Acknowledgments The authors of the present preliminary study are greatly indebted to Ben Copeland Jr., CEO of Supersod Inc. at Fort Valley for his multi-faceted contributions leading to the present study. The authors of the present paper also wish to thank the Lawn Institute for partial support of the present work.

References Allaire SE, Dufour-L’Arrivee C, Lafond JA, Lalancette R, Brodeur J (2008) Carbon dioxide emissions by urban turfgrass areas. Can J Soil Sci:529–532 Aubinet M, Vesala T, Papale D (eds) (2012) Eddy-covariance: a practical guide to measurements and data analysis. Springer. Atmospheric Sciences, Berlin Bartlett MD, James IT (2011) Are golf courses a source or sink of atmospheric CO2: a modeling approach. Proc IMechE Part P J Sports Eng Tech 225:75–83 Bremer DJ, Ham JM (2005) Measurement and partitioning of in situ carbon dioxide fluxes in turfgrasses using a pressurized chamber. Agron J 97:627–632 Dugas W, Heuer M, Mayeux H (1999) Carbon dioxide fluxes over bermudagrass, native prairie, and sorghum. Agric For Meteorol 93:121–139 Fissore C, Hobbie SE, King JY, McFadden JP, Nelson KC, Baker LA (2012) The residential landscape: fluxes of elements and the role of household decisions. Urban Ecosyst 15:1–18 Frolking SE, Bubier JL, Moore TR, Ball T, Bellisario LM, et al. (1998) Relationship between ecosystem productivity and photosynthetically active radiation for northern peatlands. Glob Biogeochem Cycles 12:115–126 Ham JM, Owensby CE, Coyne PI, Bremer DJ (1995) Fluxes of CO2 and water vapour from a prairie ecosystem exposed to ambient and elevated atmospheric CO2. Agric For Meteorol 77:73–93 Haselbauer WD, Thoms AW, Sorochan JC, Brosnan JT, Schwartz BM, Hanna WW (2012) Evaluation of experimental Bermudagrasses under simulated athletic field traffic with perennial Ryegrass overseeding. Hort Technology 22:94–98 Hull RJ (2000) Mowing, its impact on turfgrass. Turfgrass Trends 9:1–8 Kljun N, Calanca P, Rotach M, Schmid HP (2004) A simple parameterization for flux footprint predictions. Bound Layer Meteorol 112:503–523 Leclerc MY, Foken T (2014) Footprints in micrometeorology and ecology. Springer, p. 239 Leclerc MY, Thurtell GW (1990) Footprint prediction of scalar fluxes using a Markovian analysis. Bound-Layer Meteorol 52:247–258 Mariko S, Urano T, Asanuma J (2007) Effects of irrigation on CO2 and CH4 fluxes from Mongolian steppe soil. J Hydrol 333:11–123 Michaelis L, Menten ML (1913) Die kinetic der Inwertin Wirkung. Biochem Z 49:333–369 Milesi C, Running SW, Elvidge CD, Dietz JB, Tuttle BT, Nemani RR (2005) Mapping and modeling the biogeochemical cycling of turf grasses in the United States. Environ Manag 36(3):426–438

Int J Biometeorol (2017) 61:103–113 Rannik Ü, Vesala T (1999) Autoregressive filtering versus linear detrending in estimation of fluxes by the eddy covariance method. Bound-Layer Meteorol 91:259–280 Rannik Ü, Sogachev A, Foken T, Göckede M, Kljun N, Leclerc M, Vesala T (2011) BEddy-covariance handbook^. In: Footprint analysis. Springer Atmospheric Sciences, pp. 211–262 Ruimy A, Jarvis PG, Baldocchi DD, Saugier B (1995) CO@ fluxes over plant canopies and solar radiation: a review. Adv In Ecolo Reas 26:1–53 Selhorst AL, Lal R (2011) Carbon budgeting in golf course soils of Central Ohio. Urban Ecosyst 14:771–781 Townsend-Small A, Czimczik CI (2010) Carbon sequestration and greenhouse gas emissions in urban turf. Geophys Res Lett 37:L02707 Vesala T, Kljun N, Rannik U, Rinne J, Sogachev A, Markkanen T, Sabelfeld K, Foken T, Leclerc MY (2007) Flux and concentration footprint modelling: state of the art. Environ Pollut 152:653–666

113 Vickers D, Mahrt L (1997) Quality control and flux sampling problems for tower and aircraft data. J Atmos Ocean Technol 14: 512–526 Webb EK, Pearman GI, Leuning R (1980) Correction of flux measurements for density effects due to heat and water vapor transfer. Q J R Meteorol Soc 106:85–100 Wilczak JM, Oncley SP, Stage SA (2001) Sonic anemometer tilt correction algorithms. Bound-Layer Meteorol 99:127–150 Wu J, Bauer ME (2012) Estimating net primary production of turfgrass in an urban-suburban landscape with QuickBird imagery. Remote Sens 4:849–866 Zhang P, Shiping C, Zhang W, Miao H, Chen J, Han X, Lin G (2012) Biophysical regulations of NEE light response in a steppe and a cropland in Inner Mongolia. J Plant Ecol 5(2): 238–248