Implications of the spatial variability of landfill emission rates on geospatial analyses

https://doi.org/10.1016/S0956-053X(03)00102-8Get rights and content

Abstract

Accurate methods quantifying whole landfill surface flux of methane are important for regulatory and research purposes. This paper presents the results from the analysis of chamber measurements utilizing geospatial techniques [kriging and inverse distance weighting (IDW)] to arrive at an estimation of the whole landfill surface flux from the spatially distributed chamber measurement points. The difficulties in utilizing these methods will be discussed. Methane flux was determined on approximately 20 m grid spacing and variogram analysis was performed in order to model spatial structure, which was used to estimate methane flux at unsampled locations through kriging. Our analysis indicates that while the semi-variogram model showed some spatial structure, IDW was a more accurate interpolation method for this particular site. This was seen in the comparison of the resulting contour maps. IDW, coupled with surface area algorithms to extract the total area of user defined contour intervals, provides a superior estimate of the methane flux as confirmed through the methane balance. It is critical that the results of the emissions estimates be viewed in light of the whole cell methane balance; otherwise, there is no rational check and balance system to validate the results.

Introduction

There are several contributors to the increasing amount of methane in the atmosphere. These include natural wetland emissions and anthropogenic sources from rice production, enteric fermentation from ruminant livestock, and landfilling of solid wastes (Bingemer & Crutzen, 1987, Bogner et al., 1995, Bogner & Matthews, 2003, Gourlay, 1992, Neue et al., 1994). Methane gas alone contributes approximately 15% to the potential global warming estimates (OTA, 1991). Landfills have been implicated as a major source of atmospheric methane emissions (Kreileman and Bouwman, 1994), comprising about 11% of the total anthropogenic global methane contribution (Blaha et al., 1999, Boeckx et al., 1996). Current estimates based on estimates of solid waste landfilled, the quantity of methane generated, and the net methane emission, suggest that the annual global methane emission from landfills is between 14 and 40 Tg (Boeckx et al., 1996, Bogner & Matthews, 2003). However, these estimates are the results of modeling versus direct measurement. Measured methane emission rates vary over seven orders of magnitudes (0.0004–4000 g m−2 d−1) (Bogner & Spokas, 1993, Bogner et al., 1997, Czepiel et al., 1996).

Methane emissions from landfills can be controlled or mitigated. Combinations of installed gas recovery systems as well as the natural attenuation potential of various engineered covers control the rate of gaseous emissions from the landfill surface (Bogner et al., 1995, Christtoperson et al., 2000, Mosher et al., 1999). Previous field efforts have shown a significant portion (30–100%) of the methane present in the cover is oxidized to carbon dioxide by indigenous methanotrophic bacteria within the soil cover materials (Bogner et al., 1995, Christtoperson et al., 2000, Jager & Peters, 1985, Mancinelli, 1995, Whalen et al., 1990). In addition, field studies have also shown that if the methane flux is less than the methane oxidizing capacity of the landfill cover, the soil cover bacteria would oxidize atmospheric methane as it diffuses into the soil (Bogner et al., 1995).

The estimation of the whole surface emission rate can be problematic when it is measured by limited discontinuous surface chambers. The complication results from the heterogeneity of the resulting flux measurements across the surface of the landfill (Bogner et al., 1995, Cardellini et al., 2003, Jones & Nedwell, 1993, Mosher et al., 1999). The emission of methane from the surface of the landfill is a complex interaction of biological, chemical, and physical processes occurring within the landfill cover soils with all of these processes varying on different spatial and temporal scales. The spatial variation in soil permeability, air-filled porosity, methane concentration in the soil gas, moisture content, and atmospheric pressure all affect methane emission rates. Recently, it has been shown that a barometric pressure decrease of 10 millibars caused a tripling of the methane emissions from a landfill (Cziepel et al., 2003). However, this effect will be site specific. Studies have also shown that the heterogeneity in the surface flux can be related to the distribution of animal burrows in the cover soils (Giani et al., 2002). These burrows create large macropores which act as conduits for increased methane transport.

It can take several days to collect enough chamber flux measurements to describe large landfill surface areas, which could mean changes in the spatial distribution of the flux while measurements are still being collected (Mosher et al., 1999). With a lower sampling density, a regular grid pattern can provide a better estimation than random or cellular stratified sampling schemes (Wang and Qi, 1998). Reliability increased for all sampling techniques with the number of points used in the model (Wang and Qi, 1998).

Commonly the arithmetic mean multiplied by the surface area has been used in past efforts due to the fact that this provides an unbiased estimate regardless of the underlying distribution (Bogner et al., 1995, Cardellini et al., 2003, Livingston & Hutchinson, 1995). However, this technique can bias the estimate for the surface emission depending on the spatial variability and spatial extent of the higher flux regions of the surface. Often there are a few higher flux measurements (often spatial clustered), utilizing the arithmetic mean can over—or underestimate the surface emission rate since all measurements are equally weighed regardless of the “hot spot” area. Heterogeneity spanning four orders of magnitude has been measured on a single site (Pokryszka et al., 1995).

Soil moisture also has a significant controlling role for the methane emissions. Moisture contents of 15–30% (w/w) have been found to be optimal for methane oxidation activity (Boeckx et al., 1996, Giani et al., 2002). In addition, as the soil moisture increases the available pore space for gaseous transport and diffusion is reduced. Both of these factors impact the resulting methane emission. Variability in landfill cover thickness has also been shown to effect the resulting emission of methane to the atmosphere (Nozhevnikov et al., 1993), with thicker covers reducing methane flux.

Geographical information systems (GIS) have been used primarily in the site screening for landfill locations (e.g. Charnpratheep et al., 1997) and assessing the demand for solid waste disposal sites in urban cities (e.g. Leao et al., 2001). GIS has also been used for data presentation and visualization of the spatial relationship between soil gas probe readings (Moore et al., 1995), as well as to visualize the impact of landfills on the methane concentrations in urban areas, indicating that the urban landfills are a major source of elevated methane concentrations (Ito et al., 2001).

The use of geospatial models to estimate the distribution of environmental phenomenon is becoming increasingly popular (e.g. Critto et al., 2003, Gerlach et al., 2001, Leenaers et al., 1990, McBratney & Webster, 1983). Geostatistics is the term used to describe a range of statistical techniques for determining the relationship between spatially distributed values, leading to the estimation of the property at unsampled locations (Chappell, 1998). Geostatistics can interpret the fluctuations in data with respect to spatial and/or temporal variation (Olea, 1991). There are a variety of methods for representing continuous surfaces in digital form using computers (Gumbo et al., 2001). For the application here the digital elevation model (DEM) is the most useful form for the geospatial analysis (Gumbo et al., 2001). A DEM is a collection of geo-referenced elevation points with an interpolated surface. These points can be in a regular or irregular grid arrangement (Gumbo et al., 2001). To have an accurate DEM, it is necessary to acquire high enough data density to capture the features that you want to display. An overview of the geostatistical interpolation methods is given by Carusa and Quarta (1998).

The fundamental tool in the geospatial analysis is the semi-variogram, which determines the amount of spatial dependency (autocorrelation) in the data from the underlying spatial features of the variations (Burgess & Webster, 1980, Chappel, 1998, Oliver & Webster, 1987, Sorey et al., 1998, Webster & Oliver, 1990). The semi-variogram is calculated from the sampling points, and it has been recommended that at least 100 data points are needed for an accurate semi-variogram for a stationary random function (Webster and Oliver, 1992). In the real world, it is impossible to get high enough data density to fully characterize the surface emissions at every point due to practical constraints and timing. Kriging refers to the process of using the spatial dependency to predict the values of a property at unknown locations from the relationship found in the sampled locations. Kriging can be thought of as an optimal predictor (Journel and Huijbregts, 1981). The weights for the kriging analyses are derived from the semi-variogram (Oliver and Webster, 1987). Geospatial models can deal with abnormally large skewness and deviation from normal distributions (Juang et al., 2001). In addition, it has been found that a non-linear kriging model can be applied to predict concentration and volume content averages for highly skewed data sets (Kitanidis and Kuo-Fen, 1996).

There has only been limited use of geospatial techniques for estimating the resulting surface emission from landfills. Cardellini et al. (2002) states that a reliable surface emission estimate can only be accomplished through numerous measurements and a subsequent geostatistical treatment of the data. Borjesson et al. (2000) concluded that the geostatistical analysis provided a qualitative map of the surface methane flux distribution. The goal of this paper was to examine the application of geospatial statistics (kriging and IDW) to improve the quantification of methane emissions versus a simple qualitative representation.

Section snippets

Site description

The particular landfill site that was investigated in this study was the Onyx Lapouyade landfill situated near Bordeaux in France. This site has been operating since October 1996, receiving approximately 160,000 metric tons of waste per year. This site consists of two different cover configurations: (1) a final covered zone since 1998 and (2) an operating zone including temporarily covered cells with and without biogas recovery. This is shown graphically in Fig. 1. The area that was

Digital representation of landfill surface

A DEM was created for the site using the georeferenced data collected. The DEM with sample locations can be seen in Fig. 2. A spherical variogram was used to determine the kriging weights for the interpolation of the DEM. The variogram for the DEM was modeled in GS+ (Gamma Design Software, 2002) and the kriging interpolation was completed in Surfer (Golden Software, 2001) using the model parameters determined in GS+. Block kriging was chosen on a 2×2 m grid spacing and the interpolation

Conclusions and implications

The goal of this paper was to examine the potential shortcomings of utilizing geospatial methodologies in determining whole landfill emission rates. These methods offer the potential of calculating whole site emission estimates from limited point measurements, which could lead to improving overall national inventories for global landfill methane emission estimates. The major disadvantage of the chamber measurements having a small footprint also enables detailed spatial distribution studies of

Acknowledgements

Funding for this project was provided by the Centre de Recherches pour l'Environnement l'Energie et le Dechets (CREED), Limay, France as part of the current METHAN project. The content of this paper does not necessarily represent the views of this agency.

References (61)

  • P. Laville et al.

    Nitrous oxide fluxes from a fertilized maize crop using micrometeorological and chamber methods

    Agricultural and Forest Meteorology

    (1999)
  • S. Leao et al.

    Assessing the demand of solid waste disposal in urban region by urban dynamics modeling in a GIS environment

    Resources Conservation, and Recycling

    (2001)
  • H. Leenaers et al.

    Comparison of spatial prediction methods for mapping floodplain soil pollution

    Catena

    (1990)
  • A.N. Nozhevnikova et al.

    Emission of methane into the atmosphere from landfills in the former USSR

    Chemosphere

    (1993)
  • M.A. Oliver et al.

    A geostatistical investigation of the spatial variation of radon in soil

    Computers & Geosciences

    (2001)
  • X.J. Wang et al.

    The effects of sampling design on spatial structure analysis of contaminated soil

    The Science of The Total Environment

    (1998)
  • H.G. Bingemer et al.

    The production of methane from solid wastes

    Journal of Geophysical Research

    (1987)
  • J. Bogner et al.

    Kinetics of methane oxidation in a landfill cover soiltemporal variations, a whole-landfill oxidation experiment, and modeling of net-methane emissions

    Environment Science and Technology

    (1997)
  • Bogner, J., Matthews, E., 2003. Global methane emissions from landfills: new methodology and annual estimates...
  • G. Borjesson et al.

    Methane fluxes from a Swedish landfill determined by geostatistical treatment of static chamber measurements

    Environmental Science and Technology

    (2000)
  • I.M. Burgess et al.

    Optimal interpolation and isarithmic mapping of soil properties. I. The semivariogram and punctual kriging

    Journal of Soil Science

    (1980)
  • C.A. Cambardella et al.

    Field-scale variability of soil properties in Central Iowa soils

    Soil Science Society of America Journal

    (1994)
  • Cardellini, C., Chiodini, G., Frondini, F., Granieri, D., Lewicki, J., Peruzzi, L., 2003. Accumulation chamber...
  • C. Carusa et al.

    Interpolation methods comparison

    Computers Math. Applications

    (1998)
  • J. Chanton et al.

    Seasonal variation in methane oxidation in landfill cover soil as determined by an in situ stable isotope technique

    Global Biogeochemical Cycles

    (2000)
  • A. Chappel

    Using remote sensing and geostatististics to map 137Cs-derived net soil flux in south-west Niger

    Journal of Arid Environments

    (1998)
  • M. Christopherson et al.

    Methane Oxidation at low temperatures in soil exposed to landfill gas

    Journal of Environmental Quality

    (2000)
  • Coops, O., Luning, L., Oonk, H., Weenk, A. 1995. Validation of Landfill Gas Formation Models. In: Proceedings from...
  • Czepiel, P.M., Shorter, J.H., Mosher, B., Allwine, E., McManus, J.B., Harriss, R.C., Kolb, C.E., Lamb, B.K., 2003. The...
  • P. Czepiel et al.

    Quantifying the effect of oxidation on landfill methane emissions

    Journal of Geophysical Research: Atmosphere

    (1996)
  • Cited by (93)

    View all citing articles on Scopus
    View full text