Quantification of methane sources in the Barnett shale (Texas) using the Penn State WRF-Chem-FDDA realtime modeling system

Methane emissions from natural gas production areas are subject to large uncertainties at regional scales. Top-down methodologies offer an integrated approach to monitor these emissions but highly depend on the quality of the atmospheric model used to relate the surface emissions to the observed atmospheric concentrations. Using continuous measurements of in-situ CH4 and C2H6 mixing ratios from an intensive aircraft campaign over the Barnett Shale area in March 2013, we evaluated the performance of the WRF-Chem modeling system using Four Dimensional Data Assimilation (FDDA) [Deng et al., 2012b] with World Meteorological Organization (WMO) surface stations and rawindsondes, aircraft meteorological measurements (wind, temperature, humidity), and vertical profiles of the mean horizontal wind from a Doppler Lidar. We show here that Mean Absolute Errors (MAE) of the direction and speed of the wind decreased with the assimilation of the additional aircraft and wind lidar data. However, the observed spatial variability across the domain suggests that several profilers or repeated aircraft transects are required to decrease significantly the MAE over the domain. Using our five WRF-FDDA simulations coupled to a Lagrangian Particle Dispersion Model (LPDM) [Uliasz, 1994], we evaluated the concentration footprints along the flight transect and quantified the sensitivity of the location and magnitude of the footprints to the meteorological fields. We show here that the WRF-FDDA-LPDM system is capable of distinguishing between the two major contributors to methane emissions in the area, i.e. from the urban area of Dallas-Fort Worth (DFW) and from the Barnett shale gas activities.