Introduction

Landfills have been used as a convenient means to handle domestic and industrial solid wastes at low costs. Within the anaerobic conditions found in landfills, organic wastes are partially converted to landfill gas (LFG), which contains 50 to 70 % of methane (CH4), 30 to 50 % of carbon dioxide, and trace amounts of nitrogen, hydrogen sulfide, and non-methane hydrocarbons (Schroth et al. 2012). CH4 emission rates may vary over several orders of magnitude, from 0.4 × 10−3 to 4000 g m−2 d−1 (Bogner et al. 1997; Czepiel et al. 1996), depending on the age, organic content, and design of the landfill. Globally, it has been estimated that landfills account for 8 to 15 % of anthropogenic CH4 emissions (USEPA 2006a, b) while Fung et al. (1991) estimated that 40 to 60 megatons of CH4 are emitted annually.

LFG collection systems have gained general acceptance for several reasons such as government regulations; odor; and health problems, safety, and energy recovery. However, these systems are not fully efficient, and it has been estimated that they capture only between 50 and 90 % of the total LFG produced (Capaccioni et al. 2011; Themelis and Ulloa 2007). Jung et al. (2011) listed the main technical reasons for the emission of LFG into the atmosphere despite these collection systems. They include cracks, fissures, and the permeability of the landfill covers. Therefore, landfills, with or without LFG collection systems, are an important source of CH4 emissions that should be mitigated.

In developed countries, landfills are not the primary choice for waste disposal, but in existing landfills, the LFG emissions are reduced through the optimization of the collection systems and improvements in landfill practices (Bogner et al. 2008). In developing countries, landfills are still the primary choice for waste management but efforts for the implementation of cleaner methods are being made (Couth and Trois 2010). In order to meet targets for LFG emission reduction, the International Solid Waste Association (ISWA 2009) stated: “accurate measurements and quantification of greenhouse gas emissions is vital in order to set and monitor realistic reduction targets at all levels.” In order to achieve this target, several methods have been developed to detect and measure CH4 emissions from landfills, and they can be classified into three groups: (i) above-ground techniques, (ii) below-ground techniques, and (iii) ground-surface enclosure techniques. These methods have been exhaustively described by Bogner et al. (1997) and by Scheutz et al. (2009).

Above-ground techniques

are based on measurements of the CH4 concentration in the air, associated with micrometeorological or mass balance dispersion models, with or without tracer gas. These techniques allow for the determination of global landfill emissions, including those from large landfills; in addition, they can be automatized (McBain et al. 2005) and have no effect on emissions (Lohila et al. 2007). However, these techniques do not identify point sources or hotspots (Babilotte et al. 2010), require expertise (Scheutz et al. 2009), sophisticated instrumentation, and have surface constraints that limit their applications (Bogner et al. 1997).

Below-ground techniques

determine vertical fluxes from gas concentrations beneath the surface of the landfill, based on Fick’s Law. These techniques enable the determination of biogeochemical, transport, and oxidation processes (Bogner et al. 1997), but do not detect and localize point sources and are therefore not generally extrapolated to large areas.

Ground-surface enclosure techniques

are used for measuring single point fluxes (Rolston 1986). They are based on simple and direct flux measurements, within, for instance, closed chambers. They are inexpensive, simple to deploy, and interpret, and have also been validated for a variety of soils (Dalal et al. 2008). The main limitation of ground-surface methods is that they describe fluxes at specific locations and require a relatively large number of measurements before being statistically representative of global landfill emissions (Spokas et al. 2003). Thus, within reasonable time limits, ground-surface methods do not allow for the quantification of spatial variations from scale point measurements (Abichou et al. 2006), or to detect point sources or hotspots (Borjesson et al. 2000).

Several studies have shown agreement between above-ground and ground-surface methods for estimating landfill emissions (Czepiel et al. 1996; Tregoures et al. 1999; Mosher et al. 1999; Spokas et al. 2006; Lohila et al. 2007). However, in some cases, lack of correlation between these methods has also been reported. This discrepancy has been explained by either strong sources not covered by the static chambers (Borjesson et al. 2000) or by emissions from additional sources to those studied within the landfill area (Spokas et al. 2003).

Complementary to these quantification methods, reconnaissance techniques that allow for the detection of hotspots have also been developed (EA 2010). These techniques are based on the measurement of gas concentration close to the surface of the landfill, facilitating a qualitative appraisal of gas emission. However, according to Abichou et al. (2011), Giani et al. (2002), and Jones and Nedwell (1993), surface air concentration at a fixed and short distance from the soil/atmosphere interface is positively correlated with fluxes. This statement complies with the atmospheric dispersion theory (McRae et al. 1982; Holmes and Morawska 2006) and diffusion theory (Bogner and Spokas 1993). In other words, after proper calibration, flux can be quantified from simple measurements of the gas concentration close to the surface of the landfill. The latter offers a convenient method to rapidly quantify LFG fluxes together with detecting hotspots.

The objective of this study was to develop a method to determine the CH4 concentration at the ground/atmosphere interface with a surface probe. This surface probing method was calibrated with a standard closed chamber technique and then tested in a landfill with a LFG collection system.

Materials and methods

Site description and field campaigns

A municipal solid wastes disposal center was selected for field studies. For confidentiality reasons, the exact location of this landfill is not disclosed. This landfill includes a clay cover and a LFG collection system that is used to produce electricity. The studied area of the landfill was approximately 900 m long by 100 m wide, with a total surface area of approximately 90,000 m2 of even area, with sparse vegetation. This landfill was located in northeast Mexico, 500 m above sea level, and in a region with dry-warm climate, with a yearly average temperature of 22.3 °C. Two field experiment campaigns were conducted, one in December 2011 and the other in June 2012. The first campaign was dedicated to obtaining flux measurements by static chambers (SC) only, while the second campaign was dedicated to the development of the surface probing method. During both the field campaigns, there was sunny weather and wind conditions <5 m s−1, which are the meteorological conditions recommended by the USEPA for the OTM10 method (USEPA 2006a, b). During the field campaigns, atmospheric temperature, pressure, and wind velocity were measured using a portable weather station (ABH-4225, Reed, Mexico). Details of the landfills and the weather data during the campaigns are listed in Table 1.

Table 1 Landfill and experimental conditions
Full size table

CH4 detectors

CH4 concentrations were measured using two portable CH4 detectors (Extech HS-680, Sewerin, Germany) equipped with a gas sensitive semiconductor and infrared sensors, with a detection range from 0 ppm to 10 % volume CH4, and with a lower detection limit of 1 ppm above background. The detectors included an internal vacuum pump with an air flow rate of 50 ± 5 L h−1. The same CH4 detector was used for both SC and surface probing methods. CH4 detectors were laboratory calibrated before field studies using CH4-free air and a 50-ppm CH4 standard.

Methods description

Static chamber measurements were obtained using semitransparent polyethylene cylinders (volume of 1.09 × 10−2 m3 and an area in contact to the ground of 5.7 × 10−2 m2), equipped with a small fan for internal air recirculation. An external and flexible PVC skirt was used to seal the SC to the ground, thus preventing gas exchange with the exterior. Sealing was additionally ensured with two steel chains (0.5 kg each) placed on the skirt, around the SC. The detectors were placed inside the SC. CH4 concentrations inside the SC were acquired with a 0.1–1 Hz frequency, depending on the emission rate. CH4 fluxes (F, gCH4 m−2 d−1) were determined from the slope of the gas concentration increase versus time (Eq. 1); where dC/dt is the measured slope of CH4 concentration in the SC (ppm h−1), M w is the molecular weight of CH4 (16 g mol−1), P is the atmospheric pressure at measurement (atm), V SC is the SC volume (1.09 × 10−2 m3), 24 is the time conversion factor (h to d), R is the gas constant (8.2 × 10−2 atm L mol−1 K−1), T is the temperature at measurement (K), A SC is the area of the SC in contact with the ground (5.7 × 10−2 m2), and 1,000 is the conversion factor (mL to L). In the few cases where a non-linear trend was observed, the flux was calculated after Hutchinson and Mosier (1981). In all cases, fluxes were calculated using the HMR library (Pedersen 2012) in the R language (R Core Team 2012).

$$ F=\frac{\mathrm{dC}}{\mathrm{dt}}\times \frac{M_{\mathrm{w}}}{R}\times \frac{P}{T}\times \frac{V_{\mathrm{SC}}}{A_{\mathrm{SC}}}\times \frac{24}{1000} $$
(1)

The surface probing method was based on the determination of the surface concentration (Cs), which was expected to be proportional to the flux. These Cs were measured through a surface plastic probe (Sewerin, Germany) in direct contact with the ground and designed to continuously extract air from the ground surface to the detector. This probe, called a “carpet probe” by the manufacturer, was 30 cm long, 25 cm wide (an area of 7.5 × 10−2 m2), and connected to the detector by means of a 2.50 m long plastic tube. Transport time from the probe to the detector was calculated to be 2.3 ± 0.3 s, taking into account the inner volume of the plastic tube and the detector flow rate. Transport time was added to response time (t 90 < 7 s; Sewerin 2012), which yielded an average device response time of about 9.3 s.

The temporal trend of the Cs measurements, after positioning the carpet probe, was assessed. First, the Cs data were rescaled to Cs’ in such a manner that, for a given sampling point, Cs’ varied on an adimensional scale from 0 to 1. This was done according to Eq. 2, where Cs min and Cs max are the minimum and maximum measured Cs, respectively.

$$ C{s}^{\prime }=\frac{Cs-C{s}_{\min }}{C{s}_{\max }-C{s}_{\min }} $$
(2)

The surface probing method is based on a correlation between the Cs and the fluxes. To establish this correlation, both parameters were measured together at 15 locations. In each location, triplicate flux measurements were done with the SC and after each flux replicate, the Cs values were measured, each for 3 min. As described in the results section, stable Cs readings were obtained after 30 s. Thus, the replicate Cs measurements from 30 s to 3 min were averaged and used for correlation with the SC fluxes. Cs and CH4 fluxes data sets were tested for normality (Shapiro Wilks test) with R language. As detailed in the results section, the fluxes and Cs were not normally distributed, and they were therefore log converted prior to treatment. A Pearson product–moment correlation analysis between log Cs and log CH4 fluxes was also performed (R core team 2012).

CH4 flux mapping

As previously mentioned, the first campaign was dedicated only to flux measurements by the SC. The second campaign was dedicated to the development of the surface probing method. In both cases, a map of CH4 fluxes was obtained. In order to do this, the landfill area was divided into a regular grid pattern, yielding a total of 112 SC and 217 surface probing points. These numbers were above the values recommended by the Environmental Agency for monitoring landfills with an approximate surface area of 90,000 m2(EA 2010). The location at each sampling point was recorded with a global positioning system (GPSmap 76 CSX, Garmin, USA) with an absolute precision of 5 m and a relative precision of 0.85 m. At each measurement point during the first campaign, triplicate fluxes were measured by the SCs. During the second campaign, triplicate Cs measurements were done, at 30, 40, and 50 s after the surface probe was positioned. The Cs data were averaged, log transformed, and used to determine the fluxes from the correlation analysis. Fluxes were interpolated with the Surfer 11.0 software (Golden Software, USA). The selection of the best interpolation method, out of the 10 methods included in the Surfer software, was based on two criteria––the mean absolute error (MAE) and the mean bias error (MBE; Willmott and Matsuura 2006). After selecting the interpolation method, the anisotropy (ratio and angle) of the data set was determined to further minimize the MAE. The interpolation methods provided contour maps and the spatial distribution of the CH4 fluxes. The total emission from each landfill was obtained by integrating the interpolated fluxes over the spatial distribution (Borjesson et al. 2000).

Results and discussion

Flux measurements and CH4 mapping by the SCs

Figure 1 shows examples of CH4 concentration trends in the SCs observed with low, medium, and high fluxes. In most cases, the CH4 concentration increased linearly, with a correlation coefficient (R 2) >0.90. From a total of 112 triplicate flux measurements by the SCs during the first campaign, the R 2 ranged from 0.90 to 0.99 with an arithmetical mean of 0.96. The coefficients of variation (CV: standard deviation divided by the arithmetic mean of the measurements) of the triplicate flux measurements averaged 36 %. The fluxes were not normally distributed but lognormally distributed. The fluxes ranged from 0.33 to 109 gCH4 m−2 d−1 with an average of 13 ± 26 gCH4 m−2 d−1.

Fig. 1
figure 1

Example of CH4 concentration trends in SC during flux measurements

Full size image

Figure 2a shows the map of the CH4 fluxes determined by the SCs during the first campaign. This map was obtained by an inverse multiquadratic radial basis function, which was the method that best complied with the selection criteria: a minimum MAE of 16.360 gCH4 m−2 d−1 and a MBE of 0.123 gCH4 m−2 d−1. As shown, the northern section of the landfill was observed to emit more CH4. The SC technique allowed for hotspot localization as well as for the estimation of global CH4 emission of the landfill section. The global emission was estimated to be 8.1 ± 21.1 gCH4 m−2 d−1 (arithmetic mean ± standard deviation) or 30 ± 77 t ha−1 y−1, with a geometric mean of 3.0 ± 3.4 gCH4 m−2 d−1 (geometric mean ± geometric standard deviation) and a CV of 261 % for the entire landfill. It is worthwhile to mention that the elaboration of the map required two SCs, each operated by two operators over three working days for a total of 96 man-hours.

Fig. 2
figure 2

CH4 flux maps determined from a SC method and b surface probing method in northeast Mexico

Full size image

Cs measurements and surface probing method

The first measurements, made just after the carpet probe was positioned at the measurement point, showed that the Cs was changing over time. To assess this time trend, the Cs was measured at 17 locations for 120 s, and the CV trend of the Cs’ was determined (Fig. 3a). It was observed that the CV reached a steady state 30 s after positioning the carpet probe, which indicates that the Cs were no more changing over time, in excess of the standard deviation. This time was significantly longer than the response time of the surface probing device (9.3 s) and may be explained by the establishment of equilibrium between the vacuum pump, the carpet probe, the ground, and the atmosphere. Based on the sampling gas flow rate and the surface of the carpet probe, the specific sampling rate was about 650 L m−2 h−1, which indicates that the air sampled was not only coming from the space between the ground and the carpet probe but also from the edge of the carpet probe, at ground level. To confirm that the Cs reached a steady state after 30 s, the Cs was measured in one location for 1.3 h. Figure 3b shows that no time trend was perceptible over such a long period of time. This suggests that measurements done after 30 s were not the result of a transient state and that there was no exhausting effect such as the draining of CH4 absorbed in the superficial layer of the ground. From these results, it was decided to measure the Cs at each location of the landfill, 30, 40, and 50 s after the surface probe was positioned and averaged.

Fig. 3
figure 3

a CV trend of Cs’ measured in 17 locations over 120 s measurements and b example of long-term trend of Cs

Full size image

Figure 4 shows the Pearson product–moment correlation between the fluxes and the Cs measurements, including the main statistical parameters. As observed, a linear correlation between Log F and Log Cs was observed. These results confirm that the Cs can be used to measure fluxes. Log Cs were then used to estimate log fluxes in 217 points and interpolate them, as previously done with the SC method. Figure 2b shows the map obtained. Similar to the map elaborated from the SC measurements (Fig. 2a), this map was elaborated by an inverse multiquadratic radial basis function, which was the method that best complied with the selection criteria: a minimum MAE of 0.242 gCH4 m−2 d−1 and a MBE of −0.003 gCH4 m−2 d−1. Figure 2b shows that the surface probing method also allowed for hotspot localization, seen to be more concentrated in the northeast section of the landfill, as well as the determination of the global CH4 emission of the landfill section. This global emission was estimated to be 5.6 ± 10.6 gCH4 m−2 d−1, or 21 ± 39 t ha−1 y−1, with a geometric mean of 3.6 ± 2.3 gCH4 m−2 d−1, and a CV of 187 % for the landfill. This global emission is not significantly different to the global emission that was estimated from the SC map. Map 2b shows a higher resolution than map 2a mostly because the map was elaborated with more data. It should be stated that the elaboration of map 2b required the work of two operators for over 6 h for a total of 12 man-hours, lower than the workload required for the SC method. Indeed, the SC technique required 96 man-hours to establish map 2a.

Fig. 4
figure 4

Correlation between log Cs and log F. Degrees of freedom (dF), coefficient of determination (R 2), and p value

Full size image

During the field application of the surface probing method, the Cs was continuously measured, even between the measurement points. These continuous measurements allowed for direct feedback during the field work, which is of great value in order to localize hotspots. In an attempt to reduce the measurement time, the correlation between the instantaneous Cs and fluxes was tested in the 17 locations used to establish the correlation shown in Fig. 4. The correlation failed, indicating that a stabilization time of about 30 s is required after positioning the carpet probe, and that shorter times did not give results with high confidence.

Conclusions

The surface probing method presented here allowed for the quantification of CH4 emissions from a landfill with an LFG collection system. After proper calibration with a standard SC method, which required a workload of approximately 6 man-hours, the surface probing facilitated the identification of hotspots and the determination of total CH4 emissions, with an additional workload of about 0.6 man-hour per hectare. This method does not require complex or expensive equipment, since a simple and affordable portable CH4 analyzer with a sensitivity of 1 ppm is adequate. To ensure the correct operation of the method, the design of the carpet probe, which is the most sensitive piece of equipment, must guarantee a correct interface between the detector and the ground/atmosphere interface. The information generated by this technique is of potential interest to landfill operators, because it estimates where, and to what extent, the CH4 emission occurs, thus allowing for potential mitigation actions.