Skip to content


The primary objective of this research was to evaluate the accuracy of the tracer dilution method (TDM). The TDM has been employed extensively during the past decade at several landfills in the US and Europe. While best practices for TDM application, e.g., tracer location, and data analysis, e.g., filtering techniques, have been developed, the overall accuracy of the method is uncertain. In this work we extended a previous EREF-supported project by incorporating a surface flux model at the landfill surface into the publicly available Weather Research and Forecasting Model (WRF) to describe temporally varying methane emissions as a function of wind and solar radiation. In our previous work, methane emissions from a landfill were spatially variable but fixed in time. By modeling landfill methane emissions with WRF, actual emissions from the landfill are known and are compared with TDM-measured emissions determined from high-resolution WRF output data that reproduce field results. Thus, measurement error associated with the TDM can be determined, which may vary with atmospheric conditions (e.g., wind), location of tracer release, and landfill geometry.

WRF was employed with the surface flux model to capture the effects of wind and solar radiation on methane flux at two landfills: Sandtown Landfill in Delaware (US), and at a Southeastern US landfill. These landfills are of different age and significantly different topography. WRF simulation results were analyzed for representative days and for several 60- minute or 90-minute measurement periods at each landfill. TDM-measured emissions at both landfills usually overestimated actual emissions, with measurement errors averaged over 60 or 90 minutes ranging from 15 to 43% at Sandtown Landfill and -5 to 81% at the Southeastern US landfill. The higher measurement errors at the Southeastern US landfill were caused by wind-dependent emissions. During some late afternoon periods at this site, wind speed decreased rapidly on the landfill surface reducing emissions, but the downwind plumes were a mixture of old and newly emitted methane and thus did not reflect methane emissions from the landfill. If emissions were averaged over a 6-hour measurement window at the Southeastern US landfill, though, the effect of these anomalous data was reduced and measurement errors decreased to 5-26%. Thus, this study shows that when methane emissions vary from the landfill surface due to wind and solar radiation, TDM measurement error may be unusually high in some periods when emissions change rapidly. However, averaging emissions over longer measurement periods reduces the influence of these outliers.

Application of the WRF model to Sandtown Landfill and the Southeastern US landfill also indicated diurnal variations in methane emissions. At the Sandtown Landfill TDM-measured emissions varied by up to a factor of 2 over a 24-hour period, while at the Southeastern US landfill emissions varied up to a factor of 8-12 over a similar period. Emissions were smaller at night when the atmosphere was stable, and higher during the day for neutral and unstable atmospheric conditions.

To ascertain the veracity of the model-predicted variations in diurnal methane emissions, emission measurements using eddy covariance (EC) and TDM were analyzed at the Southeastern US landfill for similar time periods as the simulations. EC data indicated that methane flux was positively correlated with air temperature and wind shear velocity, and negatively correlated with both air pressure and temporal changes in air pressure. Flux was most strongly correlated with air temperature when the atmosphere was unstable, while under neutral atmosphere flux correlated most strongly with wind. These results support the surface flux parameterization model we incorporated into WRF, since emissions were strongly correlated with air temperature (solar radiation) and wind and both of these factors are included in the surface flux model. EC data also showed that methane fluxes varied diurnally by a factor of 7 to 8 on average and as much as a factor of 23 on any given day. More limited TDM data at this landfill indicated that whole-landfill methane emissions varied by up to a factor of 8 in a single day. Thus, both the EC and TDM data at this site supported the significant variation in diurnal methane flux predicted from WRF simulations at this site.

Because of diurnal variations in methane emissions, it is instructive to evaluate the bias that might occur if limited TDM-measured emissions are used to estimate average 24-hour emissions. At the Southeastern US landfill, the majority of TDM measurements (59%) were made between 12:00-18:00. If only EC data in this time period are used to estimate 24-hour emissions, the average measured emissions would exceed actual 24-hour average emissions by 55, 58 and 73% for each of the three weeks of representative EC data analyzed at this site. Thus, in addition to the small bias (overestimation) observed from 60-minute and 90-minute TDM measurements in this study, because of diurnal emission variations using limited TDM measurements would result in an even greater overestimation of actual emissions. While the results of this study are limited to these two landfills, if atmospheric conditions, i.e., wind and solar radiation, play an important role in emissions at other landfills, we expect TDM-measured emissions to also overestimate daily average emissions because of diurnal variations.

This study demonstrated the utility of atmospheric dispersion modelling for understanding measurement error associated with the TDM. While the focus of this project was the TDM, several other whole-landfill measurement techniques require sampling of well-mixed gas plumes downwind of a landfill, including aircraft-based mass balance approaches and differential absorption Lidar. Well-mixed gas plumes only exist when there is sufficient mixing of the atmosphere, which just as for the TDM limits the methods to particular times of the day when unstable or neutral stability conditions exist. Thus, the influence of diurnal emissions variability on estimates of whole-landfill methane emissions that was observed in this study is likely important for other measurement techniques too.