Update on Soil CO2 Flux Monitoring at the Illinois Basin – Decatur Project, USA

Available online at www.sciencedirect.com

ScienceDirect Energy Procedia 63 (2014) 3869 – 3880

GHGT--12

Uppdate on Soil CO2 Flux Monitoring M g at the IIllinois B Basin – D Decatur Project, P USA U Carl H. Carmana, Raandall A. Lo ocke IIa*, C Curt S. Blakkleya a

Illinnois State Geologiical Survey, Prairiie Research Institu ute, University of Illinois, Il 615 East P Peabody Drive, Chhampaign, Illinois,, 61820, USA

Abstrract The Illlinois Basin – Decatur D Project (IBDP) is a larg ge-scale carbon capture c and storrage (CCS) projeect in Decatur, IIllinois, USA demonnstrating storage of one million n metric tons off carbon dioxide (CO2) in a deeep saline reservvoir. The IBDP P Monitoring, Verifiication, and Acccounting (MVA)) program at the 0.65 km2 study site includes a ssoil CO2 flux m monitoring netwoork, measured weeklly during the gro owing season since 2009 and a high-frequency flux and atmosppheric CO2 monnitoring station. Over 11,000 flux m measurements have been colleccted from the po oint network and d over 50,000 fflux measuremennts have been ccollected at a higherr frequency from m a LI-COR Bio osciences® LI-8150 multiplexerr. No anomalies related to CO2 iinjection were iddentified. We also eexamined site-sp pecific biophysiccal relationships among soil CO O2 flux, soil tempperature, and soil moisture, andd assessed the flux ddata to better ch haracterize flux variability. Thiis extensive dataaset provides ann opportunity too characterize site-wide flux variattions in detail, an nd provides insig ghts into ways th hat measured flu uxes could be evaaluated to identiify atypical variaations. © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license © 20113 The Authors. Published by Ellsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/3.0/). Selecttion and peer-rev view under respo onsibility of GH HGT. Peer-review under responsibility of the Organizing Committee of GHGT-12 Keywoords: Illinois Basin n; IBDP; CCS; CC CUS; geologic carb bon storage; MVA A; surface monitoriing; soil CO2 flux; soil temperature; soil moisture

1. Inttroduction Thhe Midwest Geological G Seq questration Co onsortium (MG GSC), one of seven Regionnal Carbon Seequestration Partnnerships funded d by the United d States Deparrtment of Enerrgy (DOE), is conducting a llarge-scale carb rbon capture and sstorage projectt in Decatur, Illinois, USA, known as the Illinois Basin – Decatur Prooject (IBDP). The project objecctive is to dem monstrate storag ge of one milliion tonnes of carbon c dioxidee (CO2) in a deeep saline reseervoir in the

* Coorresponding autho or. Tel.: +1-217-33 33-3866; fax: +1-2 217-333-2830. E-m mail address: [email protected]

1876-6102 © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/). Peer-review under responsibility of the Organizing Committee of GHGT-12 doi:10.1016/j.egypro.2014.11.417

3870

Carl H. Carman et al. / Energy Procedia 63 (2014) 3869 – 3880

Illinois Basin. Carbon dioxide injection began on November 17, 2011, at an average rate of 1,000 metric tons per day, and is scheduled to conclude in November 2014. The IBDP employs an extensive environmental Monitoring Verification and Accounting (MVA) program consisting of near-surface and subsurface monitoring to establish baseline conditions prior to injection, as well as monitoring throughout the three-year injection period, and a scheduled three-year post injection period. Over twenty near-surface and subsurface monitoring techniques have been implemented at IBDP, including soil flux monitoring. The soil flux monitoring network consists of manual and automatic measurements at fixed points for characterization of spatial flux variations over the 0.65 km2 study site, and of temporal variations on diurnal to annual time scales. This paper presents the results of soil CO2 flux monitoring at the IBDP site and describes the spatial and temporal patterns observed. 2. Methods Soil CO2 flux is the rate at which CO2 crosses the soil surface-atmosphere boundary. It is primarily driven by exudation of carbon compounds into the rhizosphere. There, mycorrhizal fungi and bacteria oxidize these organic compounds to produce energy, water, and CO2. The rates of these processes are determined by soil temperature, soil moisture, and available soil carbon [1, 2]. Therefore it is important to monitor these biophysical properties and their impact on observed fluxes, to better quantify variability and ascertain whether or not CO2 has migrated from the injection reservoir to the biosphere [3]. At the IBDP site, we employ a closed chamber accumulation method with additional concurrent monitoring of soil temperature and soil moisture. 2.1. Soil flux monitoring network A network of 107 polyvinylchloride (PVC) rings distributed at 75 locations has been sampled weekly during the growing season since July 2009. Rings were placed in a rough grid, but existing infrastructure limited ring installations to the south and east of the injection well (CCS1) (Fig. 1). In the area to the north and west of the injection well (CCS1), rings were spaced about 75 m apart. Closer to the injection well, rings partially encircle CCS1 and were spaced between 10 and 30 m apart. All rings in the network were reset and leveled at the beginning of each field season and anytime a disturbance occurred that effected the ring position. Data for this paper cover the period when monitoring began (July 2009) to the end of the monitoring season in December 2013. Meteorological temperature and precipitation data from the KDEC weather station at the Decatur Airport in Decatur, Illinois located 5 km from the site were used for the period from December 2008 through December 2013. An eddy covariance system was also used to collect meteorological data and soil temperatures at a single location at the site. Natural Resources Conservation Service (NRCS) soil surveys indicate soils at the site are mostly silt loams and silty clay loams [4]. However, land use changes and industrial activities have likely significantly changed the original soil conditions at some locations and in some cases completely removed or buried soil horizons. The site elevation varies by 5.5 m (18 ft) across the study site, from 202 m to 207.5 m above sea level. The flux monitoring network contains three ring treatments: (1) 75 bare soil, shallow-depth rings, (2) 27 natural soil, shallow-depth rings, and (3) 5 bare soil, deep-depth rings. Shallow and deep rings were driven about 8 and 46 cm into the ground, respectively. Bare ring types are regularly treated with herbicide to minimize vegetal effects within a 0.5 m radius of the ring, while natural rings are maintained only to prevent plant growth from impeding flux measurements. The bare-shallow rings have the highest signal to noise ratio of the three ring types and are expected to be the most sensitive to a leakage signal, were one to occur [5], because of minimized contributions to flux from root respiration and microbial activity. For this paper, we focused only on treatment types #1 and #2.

3871

Carl H. Carman et al. / Energy Procedia 63 (2014) 3869 – 3880

Author namee / Energy Procediia 00 (2013) 000––000

3

Fig. 1. Map of IBDP study area. Includes soil s flux point mon nitoring network, multiplexer m locatioon, eddy covariancce tower, injectionn well (CCS1), verificcation well (VW1), geophone monito oring (GM1), and CO C 2 transfer pipelline.

2.2. F Flux monitorin ng 2.2.1. Point monitoring he growing seaason, which is typically from m April to Deccember. The Sooil fluxes weree measured weeekly during th measurements weree generally tak ken between 07:00 0 and 18:0 00 hours. A LII-COR Bioscieences® LI-81000A singlember, portable soil flux system with an aux xiliary sensor interface was used to measuure fluxes. Thee LI-8100A cham utilizzes a closed-chaamber accumu ulation method with an infrareed CO2/H2O annalyzer to deteermine CO2 conncentrations and linear and expo onential regresssions to calcullate fluxes [6].. For the first ttwo years of m monitoring, co--located soil oil moisture measurements m were w collected d sporadically with flux meaasurements. Beginning in temperature and so Auguust 2012, ancilllary soil tempeerature and soill moisture data have been moore routinely coollected. Fluux measuremeent durations were w 120 seconds, per the manufacturer’s m recommendattions for low tto moderate fluxees [6], and flux estimates werre field checked d for anomalou us values. All ddata subsequenntly underwentt a thorough qualitty control rev view. Quality control criterria included: (1) comparisoon of written and electronnic data for consiistency, (2) ev valuation of recent r flux vaalues versus historical h data,, (3) resolvinng inconsistentt individual measurements and their calculatted regression n curves, and (4) using maanufacturer reccommendationns to assess measurement qualitty. Data that diid not pass the quality controll review were nnot used in datta analysis. ncy monitoring g method 2.2.2. High-frequen Hiigh-frequency soil CO2 flux monitoring ussed a LI-COR R Biosciences® ® LI-8100A innfrared CO2/H2O analyzer with a LI-COR Bio osciences® LI-8150 multipleexer. Measurem ments were coollected every thirty minutess from eight

3872

Carl H. Carman et al. / Energy Procedia 63 (2014) 3869 – 3880

ports: four closed accumulation chambers (two bare shallow and two natural shallow rings) and four atmospheric sampling ports at 9, 55, 168, and 243 cm above ground surface. Soil temperature and moisture measurements were collected at each of the four accumulation chambers. At the beginning of the project, the multiplexer was deployed only for short periods of time. In 2012, the multiplexer was relocated to its current location, and deployed for substantially longer periods of time with more than 100,000 total flux measurements collected during 2012 and 2013. All multiplexer data have been reviewed and censored based on quality control review procedures similar to those used for the point measurement network. 3. Results and discussion 3.1. Data summary A total of 11,281 fluxes that have met the project QC criteria were measured at the rings in the IBDP soil flux point network between June 2009 and December 2013. The flux data were grouped by season with spring, summer and fall periods representing March through May, June through August, and September through November, respectively. Generally fluxes were not measured from December through March because of frozen or saturated soil conditions. Measurements from the standard bare-shallow ring type account for 70% of the data, and measurements from the other two ring types, natural shallow and bare deep, account for 26 and 4 % of the data, respectively. Average soil CO2 fluxes were calculated for the bare and natural rings since flux monitoring began (Table 1). The average soil CO2 flux at natural rings (4.4 —mol m-2 s-1, n = 3164), is 2.3 times greater than at bare rings (1.9 —mol m-2 s-1, n = 7975), because natural rings have relatively unrestricted plant and root respiration and soil microbial activity, while the bare ring herbicide treatments significantly reduced these effects. Generally, the bare rings’ fluxes (STD = 1.8) were less variable than the natural rings’ (STD = 3.5). These data corroborate the previous assessment that bare rings are a preferred installation type for CO2 leak detection because of their greater sensitivity to detect flux anomalies. Table 1. Yearly average soil CO2 fluxes and flux differences (in —mol m-2 s-1) for bare and natural ring types. Measurements in 2011 after November 17 were added to the 2012 average.

Bare ring

n=

Natural ring

n=

Flux difference (natural–bare)

Start Date

End Date

2009

2.0

1057

2.7

245

0.7

6/24/2009

12/22/2009

2010

2.3

1224

5.5

481

3.2

4/1/2010

11/10/2010

2011

2.2

1214

4.6

575

2.4

5/11/2011

11/17/2011

2012

1.5

2434

4.1

917

2.6

11/28/2011

11/28/2012

2013

1.7

2056

5.1

767

3.6

5/7/2013

12/4/2013

Averages

1.9

7985

4.4

2985

2.5

6/24/2009

12/4/2013

Minimum, maximum, and average fluxes for the 75 bare soil rings for each sampling event were variable, but in general follow expected seasonal trends (Fig. 2). Minimum fluxes were between 0.0 and 1.3 —mol m-2 s-1 (range = 1.3 —mol m-2 s-1) and average fluxes were between 0.2 and 4.8 —mol m-2 s-1 (range = 4.6 —mol m-2 s-1). However, maximum flux values varied more widely from 0.5 to 27.6 —mol m-2 s-1 (range = 27.1 —mol m-2 s-1). Overall, eighty percent of flux values were below 2.2 —mol m-2 s-1. A box and whisker plot (Fig. 3) of 6,020 flux measurements shows the seasonal data distributions. Spring and summer showed the highest flux variability as expected, because these are the most active portions of the growing season. Fall had slightly less flux variability as compared to spring, and winter had the lowest flux variability, as would be expected because of generally low wintertime fluxes observed at IBDP and elsewhere [7]. In the context of leak detection, outliers with anomalously high values would be the most relevant to investigate to determine their source from natural variability, anthropogenic disturbance, leak

Carl H. Carman et al. / Energy Procedia 63 (2014) 3869 – 3880

preseence, or a comb bination of tho ose factors. A quantile-quanti q ile probability plot of all fluxxes indicated a log-normal distriibution of the data. d

Fig. 2. Aveerage, minimum, and a maximum flux xes from 2009 throough 2013 at IBDP P.

Fig. 3. Box and whisker plot of log transfo ormed CO2 fluxes (—mol m-2 s-1), gro ouped by spring (M March, April, and M May), summer (Juune, July, and Augusst), fall (Septemberr, October and Nov vember), and wintter (December, Jan nuary, and Februarry). Sample size = 6,020.

3873

3874

Carl H. Carman et al. / Energy Procedia 63 (2014) 3869 – 3880

3.2. Biophysical effects 3.2.1. Precipitation responses Fluxes were compared to average daily temperature and 30-day moving average precipitation observations from the Decatur Airport (Fig. 4). Soil flux at the IBDP site follows seasonal average temperature fluctuations: average fluxes were less than 1.0 —mol m-2 s-1 in the winter, but in spring increased rapidly and every year peaked with fluxes exceeding 3.5 —mol m-2 s-1 in either the second half of June or first half of July. They subsequently decreased steadily until reaching their lowest observed values in late November. Average summer fluxes for bare rings were calculated for each year from 2009 to 2013 (Table 2). Because of the range of observed precipitation conditions, flux variations due to precipitation differences from year to year and season to season have been well quantified. In 2012, the study site experienced the third most severe drought on record and received only 3.9 cm of precipitation during the summer (compared to an average summer precipitation of 8.0 cm). Average fluxes in the summer of 2012 were approximately 0.7 —mol m-2 s-1 (40%) to 1.6 —mol m-2 s-1 (30%) lower than fluxes measured during the summers of 2011 and 2013, respectively. Lower fluxes measured during the drought are directly related to soil moisture deficits and decreased plant growth and vigor, which can reduce total soil respiration [8, 9]. Conversely, the summer of 2010 received 14.3 cm of precipitation, and average fluxes were 20% higher than during the summers of 2009 and 2011. 50

30 Temperature SoilFlux

25

Precipitation

40

20

35 30

15

25 10

20 15

SoilCO2 Flux(ђmolmͲ2sͲ1)

AirTemperature(°C)andPrecipitation(in.)

45

5

10 0 5 Ͳ5

0 5/17/2009

12/3/2009

6/21/2010

1/7/2011

7/26/2011

2/11/2012

8/29/2012

3/17/2013

10/3/2013

Fig. 4. IBDP soil CO2 flux values measured at bare shallow rings, with temperature and precipitation data from the Decatur Airport. Precipitation is scaled up 50× and is a 30-day moving average.

3875

Carl H. Carman et al. / Energy Procedia 63 (2014) 3869 – 3880 Table 2. Average summer soil fluxes observed at bare soil, shallow-depth rings during June, July, and August of 2009 to 2013. Temperature and precipitation are from the Decatur Airport. Average summer fluxes Year

(ђmol m-2 s-1)

n=

Cumulative summer precipitation (cm.)

2009

2.9

530

8.1

22

2010

3.5

541

14.3

26

2011

2.9

481

7.0

25

2012

1.9

1002

3.9

25

2013

2.6

907

7.0

23

Average temperature (°C)

A LI-COR Biosciences® LI-8150 multiplexer was deployed four times (Table 3) to evaluate diurnal trends in soil fluxes. During these periods, a total of 53,426 atmospheric measurements and 53,430 soil flux measurements were collected. These data were used to characterize long-term trends, as well as effects of significant precipitation events on diurnal flux cycles. Moderate to heavy precipitation events can reduce the air-filled volume of soils. As soil pores become water-filled, CO2 migration in the soil is impacted by reduced advective flow of CO2 and increasing diffusive flow of water through the pores, resulting in decreased soil-atmospheric fluxes [8]. When soils become increasingly saturated, soil respiration can be reduced because microbial activity is inhibited [8]. Then, as soils become less saturated, microbial and plant activity increases rapidly, resulting in an accumulation of CO2 in the soil, which is released (Fig. 5) when soil pathways become unsaturated [10]. Table 3. Periods of LI-COR Biosciences® LI-8150 multiplexer deployment at its current location. Deployment periods June 6, 2012 – July 25, 2012 October 23, 2012 – January 8, 2013 August 9, 2013 – September 25, 2013 October 2, 2013 – November 8, 2013

Fluxes are most highly correlated with soil temperature, but low and high soil moisture can significantly disrupt flux patterns [8]. However, soil temperature data have not always been collected at the IBDP site. When soil temperature data have been unavailable, air temperature data has been used as a proxy for soil temperature. IBDP flux patterns are lagged when compared to air temperature (Fig. 5), due to the thermal diffusivity differences between soil and air [11]. On June 16 at about 22:00, a precipitation event began and about 0.9 cm, or about 20% of the total precipitation that fell in the summer of 2012, occurred (Table 2); precipitation fell for over four hours (80% in the first hour). The pattern follows known flux responses to precipitation events [9].The first wetting of the soil increased microbial respiration, which caused an initial spike in the soil flux. Ambient atmospheric temperature dropped, and combined with subsequent soil saturation, caused soil flux to decrease from 8.8 —mol m-2 s-1 to 1.5 —mol m-2 s-1. Then moisture levels decreased below saturation and temperature stabilized and then increased. Subsequently, soil respiration activity increased, which caused flux to increase from a low of 1.5 —mol m-2 s-1 at 00:30 to 11.5 —mol m-2 s-1 by 13:30 on June 17. However, this was the only significant precipitation event between June 1st and July 31st, so soil moisture content quickly returned to normal, microbial respiration of organic matter diminished, and flux resumed the typical diurnal pattern observed at the IBDP and previous studies [12]. Soil temperature and moisture are known to significantly influence fluxes, so accurately characterizing these relationships at IBDP have helped the project to evaluate observed variabilities in the context of leak detection.

3876

Carl H. Carman et al. / Energy Procedia 63 (2014) 3869 – 3880

50

AirTemperature

PrecipitationEvent

Linearflux

14

45 12

10

Temperature(°C)

35 30

8

25 6

20 15

4

LinearsoilCO2 flux(ђmolmͲ2sͲ1)

40

10 2 5 0

0 6/14/2012

6/15/2012

6/16/2012

6/17/2012

6/18/2012

6/19/2012

6/20/2012

6/21/2012

Fig. 5. Soil flux response to a 0.9 cm precipitation event on the evening of June 16, 2012.

3.2.2. Moisture and temperature dynamics In 2013, soil moisture and soil temperature data were collected in conjunction with each soil CO2 flux point measurement. Fluxes, soil temperatures, and soil moistures at the IBDP site exhibit very similar relationships to those observed by Longdoz et al. [13], and Howard and Howard [8]. Weekly averaged soil moisture and soil temperature data for bare-shallow rings showed a negative correlation (r = –0.74, r2 = 0.55, n = 28), which suggests that soil temperature and soil moisture values were moderately correlated. In environments similar to the IBDP site, soil CO2 flux correlates primarily with soil temperature, and to a lesser degree with soil moisture [14]. However, the flux-moisture relationship is highly dependent on how near soil moisture values are to either the wilting point or saturation point [15]. Flux-temperature and flux-moisture correlations were examined using 30% soil moisture as a threshold for assessment of temperature/moisture covariance based on previous studies [16, 17]. Fluxes were regressed with soil temperature and moisture, and further subdivided by ring type to examine the variation in flux signatures between bare and natural ring types (Fig. 6). Above 30% soil moisture, bare and natural ring types displayed a moderate relationship between flux and temperature (R2 = 0.50). Below 30% moisture, bare rings showed a weaker relationship between flux and temperature (R2 = 0.32), and natural rings showed no correlation (R2 = 0.05). Bare and natural ring types showed no relationship between flux and soil moisture when soil moisture was above 30%, and very low correlations (R2 = 0.13 and 0.22, respectively) when soil moisture was below 30%. Weaker correlations were likely caused by spatial variability in flux behavior, which compounded the natural variability of flux observations at individual locations. The biophysical relationships described above have been important in developing a site specific understanding of soil CO2 flux dynamics.

3877

Carl H. Carman et al. / Energy Procedia 63 (2014) 3869 – 3880

BareShallow 1.00

R2

Fluxvs.SoilTemperature

NaturalShallow Fluxvs.SoilMoisture

1.00

0.80

0.80

0.60

0.60

0.40

0.40

0.20

0.20

0.00

0.00 Moisture<30%

Moistureш30%

Moisture<30%

Moistureш30%

Fig. 6. Averaged linear regression coefficients of determination, as they relate to a 30% soil moisture threshold, for IBDP soil flux rings.

3.3. Monitoring challenges Several challenges have been encountered during the nearly five years of monitoring. On occasion, data collection was interrupted or prevented by equipment problems (e.g., loss of power), or damage resulting from severe weather, animals, or other site activities. Where possible, those situations were mitigated by system modifications or repairs. In some cases, erroneous values were measured due to a failure of the accumulation chamber to close and seal fully, which resulted in atmospheric mixing during the measurement sequence. The data review procedure was able to identify those situations and the erroneous data were removed. Staff turnover has also been a factor in maintaining consistent data collection, management and review procedures, and in some cases, it has resulted in minor to moderate data gaps. To prevent additional staffing related issues, written procedures were developed for data collection, management, and review. Natural flux variability is a very significant factor in assessing data for leakage signatures. A leak could have a range of characteristics relating to its rate, duration, and areal extent, and natural flux variability could significantly mask leakage if it were to occur. To address the challenges of leakage identification, characterization of field site heterogeneities is essential and the flux measurements need to be assessed from multiple temporal and spatial perspectives. 3.4. Pre-injection and injection period fluxes Soil CO2 fluxes from bare-shallow rings during the pre- and syn-injection periods were compared to determine if significant variability has occurred. At the coarsest level, pre-injection fluxes were compared with those measured after November 17, 2011, when injection began. Yearly site-wide flux averages (Table 1) were similar between the pre-injection (2.2 —mol m-2 s-1) and injection periods (1.6 —mol m-2 s-1). The initial late fall and winter fluxes of 2011, which were very low, and the reduced fluxes during the drought in 2012 were significant factors contributing to the lower average flux of the injection period. For a closer look at seasonal patterns and flux variability across the site, spatial analysis has been performed with GIS software. A description of the GIS analysis workflow and approach is provided in Korose et al [18]. In general, flux observations were gridded and interpolated to produce site flux maps on different time scales (e.g., weekly, monthly, seasonally, yearly) for the purpose of data exploration, visualization, and assessment. Fig. 7 shows an example of the automated output for the summer and fall seasons from 2009 through 2013. The fall fluxes are lower than the summer fluxes, and the impact of the 2012 drought can most clearly be seen in the summer 2012 map showing lower fluxes site-wide. Also of note are the fall 2011 and summer 2013 maps, which show transient periods of relatively higher fluxes at rings 5H and 11I, respectively. The cause of those values is not expected to be related to injection, but rather natural physical characteristics or construction disturbances at each of those locations which

3878

Carl H. Carman et al. / Energy Procedia 63 (2014) 3869 – 3880

havee promoted hiigher natural fluxes. f Ring 5H 5 is on top of an artificiaal berm, consstructed in 20111, and that consstruction disturrbance is expeccted to be the cause c of the hig gher observed ffluxes. Weeklyy flux maps weere reviewed and also indicate that t the higher fluxes all occcurred prior to injection. Rinng 11I lies alonng a drainage ditch, which y wet long into o the 2013 mon nitoring/growin ng season, prevventing effectiive herbicide trreatment and remaained atypically allow wing plants to o grow relativ vely unrestrictted. After reg gular herbicidee treatments ccould be resuumed, fluxes decreased to norm mal in fall 2013 3 (Fig. 7). Alsso note that oth her localized hhigher fluxes hhave been obsserved in the g., summer 2009, summer 2010). Based on these currennt assessments,, no anomalouus variability seasoonal maps (e.g was identified in th he soil CO2 flu uxes that were not n attributablee to natural varriability or connstruction activvity. In other o not indicate a leak has occu urred. wordds, the current assessments do

Fig. 77. Bare-shallow rin ng flux averages fo or summer and falll from 2009 throug gh 2013. Note: all relatively higher llocalized fluxes arre attributed to naturaal physical conditiions or constructio on disturbances.

4. Conclusions 4.1. Flux evaluatio on Soil CO2 fluxess have been meeasured on a weekly w basis siince 2009, at oover 100 locatiions at the 0.65 km2 IBDP site. Pre-injection (baseline) ( fluxes were measu ured from June 24, 2009, throough Novembeer 17, 2011, whhen injection begaan. Weekly mo onitoring throu ugh the injectio on phase has continued c to thhe present. Thee greatest obstaacle in longterm m (years) of so oil flux monittoring at the IBDP site haas been staff tturnover, whicch hinders consistency in meassurement proccedures, equip pment mainten nance, and daata completenness. Additionnally, power lloss, animal interrference, extreeme weather, or material obstructions o caan cause equippment or meaasurement faillures. These challlenges have beeen addressed by b site mainten nance and by more m thoroughlyy documentingg monitoring prrocedures. E Evaluation of soil CO2 flux xes for leakag ge monitoring is further coomplicated byy their naturall variability. Com mparison of nattural soil to herrbicide-treated soil at monito oring locations suggests that ttreated soils, w where natural vegeetation is elimiinated in the measurement m arrea, resulted in n the smallest ssignal to noise ratio, and is liikely a more sensitive treatmentt to detect a CO C 2 leak at a CCS C site, shou uld it occur. A Although soil ffluxes follow sseasonal and ndividual flux xes are greatly y affected by current c soil coonditions, notaably soil tempperature, soil diurnnal patterns, in moissture, recency of precipitatio on events, and physical ring disturbances. For effective lleak detection,, we suggest CCS S flux monitoring programss should take all of these influences innto consideratiion, and suppplement flux

Carl H. Carman et al. / Energy Procedia 63 (2014) 3869 – 3880

observations with concurrent soil temperature and moisture measurements. In addition, soil properties such as bulk density, particle density, porosity, and organic carbon content will likely help further evaluate soil CO2 flux at IBDP. Review of flux maps on multiple time scales suggests no CO2 leakage from the injection reservoir to the biosphere at the IBDP site. Additional more detailed graphical and statistical assessments are now being conducted. 4.2. Key correlations Soil temperature and moisture correlate significantly with soil CO2 flux, but do not completely predict fluxes. Soil temperature alone is significantly correlated with soil CO2 flux, typically accounting for 40% to 60% of flux variations, comparable with previous studies [13]. Soil moisture was also correlated with soil CO2 fluxes at the IBDP site, but this relationship is highly dependent on soil moisture with regards to the degree of soil saturation. Preliminary multiple regressions of log-transformed soil temperature and soil moisture against CO2 flux account for 60% to 70% of flux variability (p < 0.05). This suggests that a sizeable portion of soil CO2 efflux at the IBDP site is determined by other soil properties, likely organic matter content and porosity [19]. Therefore, accurate interpretation of soil CO2 fluxes at a project site requires baseline and ancillary soil data to determine spatial variations in soil type and available carbon. Flux networks consisting primarily of bare soil, shallow-depth rings are less susceptible to extraneous fluxes associated with plant processes, and therefore theoretically would provide more effective detection of potential leaks than ring types more influenced by surface vegetation. 4.3. Future work Monitoring of the soil CO2 flux network at the IBDP site will continue for at least three years post-CO2 injection and ancillary soil data will be collected. Additional soil characterization is planned to more completely document current soil type(s) and properties. We also plan to test methods of estimating historical soil temperatures and moistures at IBDP. These historical data will enable us to better characterize baseline flux variability and potentially build a numerical model to estimate typical fluxes at each sampling location, using soil moisture, soil temperature, and available carbon [1, 18]. We will also be able to better evaluate the effects, if any, of physical anthropogenic site disturbances. Acknowledgements The authors would like to thank Damon Garner (database development), Chris Korose (figure development), Ivan Krapac (initial network design and paper review), Stephen Picek (data review), and Alex Fine (data review) for their contributions. Chris Patterson, Elizabeth Curtis-Robinson, Nick Chin, and Jacquelyn Hurry collected data used in this paper. William Bruns helped with installation and regular maintenance of the flux network. The Midwest Geological Sequestration Consortium is funded by the U.S. Department of Energy through the National Energy Technology Laboratory via the Regional Carbon Sequestration Partnership Program (contract number DE-FC2605NT42588) and by a cost share agreement with the Illinois Department of Commerce and Economic Opportunity, Office of Coal Development through the Illinois Clean Coal Institute. References [1] Beaubien SE, Jones DG, Gal F, Barkwith AKAP, Braibant G, Baubron J-C, Ciotoli G, Graziani S, Lister TR, Lombardi S, Michel K, Quattrocchi F, Strutt MH. Monitoring of Near-surface Gas Geochemistry at the Weyburn, Canada, CO2-EOR Site, 2001–2011. International Journal of Greenhouse Gas Control 2013; 16: S236–262. [2] Vargas R, Baldocchi DD, Allen MF, Bahn M, Black TA, Collins SL, Yuste JC, Hirano T, Jassal RS, Pumpanen J, Tang J. Looking Deeper into the Soil: Biophysical Controls and Seasonal Lags of Soil COproduction and Efflux. Ecological Applications 2010; 20.6: 1569–582. [3] Brydie J, Faught B, Olson M, Underwood A, Drozdowski B. The Laboratory Simulation and Field Verification of Seasonal Soil-respired CO2 Flux at a Proposed CCS Project Site. Energy Procedia 2013; 37: 4041–4048.

3879

3880

Carl H. Carman et al. / Energy Procedia 63 (2014) 3869 – 3880 [4] Soil Survey Staff, Natural Resources Conservation Service, United States Department of Agriculture. Web Soil Survey. Available online at http://websoilsurvey.nrcs.usda.gov/. Accessed 04/22/2014. [5] Locke II RA, , Krapac IG, Lewicki JL, and Curtis-Robinson E. Characterizing Near-surface CO2 Conditions before Injection - Perspectives from a CCS Project in the Illinois Basin, USA. Energy Procedia 2010; 4: 3306–3313. [6] LI-COR Biosciences. LI-8100 Automated soil CO2 flux system and LI-8150 multiplexer. 4th ed. Instruction manual; 2007. p. 52–63. [7] Winston GC, Sundquist ET, Stephen BB, Trumbore SE. Winter CO2 Fluxes in a Boreal Forest. Journal of Geophysical Research 1997; 102.D24: 28795–28804. [8] Howard DM, Howard PJA. Relationships between CO2 Evolution, Moisture Content and Temperature for a Range of Soil Types. Soil Biology and Biochemistry 1993; 25.11: 1537–1546. [9] Sponseller RA. Precipitation Pulses and Soil CO2 flux in a Sonoran Desert Ecosystem. Global Change Biology 2007; 13.2: 426–436. [10] Jin Z, DongY-S, Qi Y-C, and Domroes M. Precipitation Pulses and Soil CO2 Emission in Desert Shrubland of Artemisia Ordosica on the Ordos Plateau of Inner Mongolia, China. Pedosphere 2009; 19.6: 799–807. [11] Wijk WRV. Physics of Plant Environment. Amsterdam: North-Holland Pub.; 1963. [12] Borken W, Davidson EA, Savage K, Gaudinski J, Trumbore SE. Drying and Wetting Effects on Carbon Dioxide Release from Organic Horizons. Soil Science Society of America Journal 2003; 67.6: 1888. [13] Longdoz B, Yernaux M, Aubinet M. Soil CO2 Efflux Measurements in a Mixed Forest: Impact of Chamber Disturbances, Spatial Variability and Seasonal Evolution. Global Change Biology 2000; 6.8: 907–917. [14] Wang Y, Amundson R, Trumbore S. The Impact of Land Use Change on C Turnover in Soils. Global Biogeochemical Cycles 1999; 13.1: 47–57. [15] Solomon DK, Cerling TE.. The Annual Carbon Dioxide Cycle in a Montane Soil: Observations, Modeling, and Implications for Weathering. Water Resources Research 1987; 23.12: 2257. [16] Jasek A, Zimnoch M, Gorczyca Z, Smula E, Rozanski K. Seasonal Variability of Soil CO2 Flux and Its Carbon Isotope Composition in Krakow Urban Area, Southern Poland. (n.d.): n. pag. PubMed. Web. 24 Apr. 2014. . [17] Volo TJ, Vivoni ER, Martin CA, Earl S, Ruddell BL. Modeling Soil Moisture, Water Partitioning and Plant Water Stress under Irrigated Conditions in Desert Urban Areas. Ecohydrology 2013. [18] Korose CP, Locke II RA, Blakley CS, Carman CH. Integration of near-surface monitoring information using ArcGIS at the Illinois Basin – Decatur Project, USA. Energy Procedia 2014 (this issue). [19] Shi, P-L, Zhang X-Z, Zhong Z-M, Ouyang H. Diurnal and Seasonal Variability of Soil CO2 Efflux in a Cropland Ecosystem on the Tibetan Plateau. Agricultural and Forest Meteorology 2006; 137.3-4: 220–233.