Investigation on the Effects of Boundary Layer Parameterization and Model Grid Size on Large-Scale High-Resolution Identification of Emission Sources of Arctic Black Carbon

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 7 January 2015
Meng-Dawn Cheng, ORNL, Oak Ridge, TN

(For the 95th AMS Annual Meeting the 17th Conference on Atmospheric Chemistry


Meng-Dawn Cheng1 & Erik D. Kabela2

1 Environmental Sciences Division

2 Nuclear Security and Isotope Technology Division

Oak Ridge National Laboratory

Oak Ridge, TN 37831-6036



Identifying emission sources remotely is of interest to global pollution research and international treaty for mitigation of climate change. It is non-trivial to obtain source data for black carbon emissions in Russia, for example[1]. Unknown geographical locations of emission sources make it difficult to perform high-fidelity source-receptor analysis[2] using models that require a priori source information. Trajectory is a useful tool for many source-receptor analysis including the Potential Source Contribution Function (PSCF) model2,[3],[4]. The PSCF model derives a map showing the probability of grid cells being emission sources based on the ambient chemistry data and backward trajectory associated with the same time period where the chemistry data were taken. Theoretically, if the PSCF were performed on a single chemical species; e.g., black carbon, the map would reveal the geographical locations of the potential emission sources. The PSCF resolution in the prior application2 was on a 2.5 by 2.5 grid domain for the hemisphere from 30N to the Pole.

In this study, we investigated the impacts of grid resolution and terrain, orographic processes, on the long-range source identification based on PSCF with back trajectories calculated by the HYSPLIT 4 model[5]. The orographic processes were parameterized by two Planetary Boundary Layer (PBL) schemes; the MYJ and YUS schemes, at two domain resolutions within the Weather Research and Forecasting (WRF) model (Version 3.6)[6]. Both variations in WRF were used to generate input data for HYSPLIT 4. These are two popular PBL schemes, with the YUS the default scheme for WRF. The two WRF domain resolutions were 83 km and 27 km, in contrast to the standard grid resolution of the NOAA/NCAR Reanalysis, 2.5 by 2.5, which is about 250 km by 250 km. The use of WRF wind field also has impacts on the vertical motion of a trajectory in addition to its horizontal movement, which could resolve the influence of regional and local meteorological processes such as jet streaks, blocking patterns, Rossby waves, and terrain-induced convection on the trajectory, which leads to impacts on the PSCF modeling result.

The following figures show examples of 10-day back trajectories calculated by HYSPLIT 4 using three wind fields; i.e., the Reanalysis, WRF-83km, and WRF-27km for January 31, 2000 1200 UTC (left panel) and 0600 UTC (right panel). The red is the Reanalysis, while the green and blue are 83km and 27km, respectively. Significant difference in the spatial traverse (coverage and direction) of the three trajectories can be seen in the figures. The blue and green trajectories appear to respond to a localized or regional process in Siberia, which altered its direction, while the small-scale phenomena did not appear in the Reanalysis data and thus not reflected in the trajectory.

The PSCF model takes an ensemble view of a large number of trajectories over the entire year of 20002 to reconstruct a potential source emission map that uses a total of 1,460 trajectories and 350,400 trajectory endpoints. If a large percentage of the 1,460 trajectories of an input data set is significantly different from that of the others, we hypothesize that the PSCF maps would be different. At what fraction of the trajectory data the PSCF would lead to a completely different forensic conclusion is an ongoing evaluation and the answer will advise the statistic robustness of the PSCF method. The result would also permit an improved understand of trajectory science and its use in PSCF in general. Finally, one of the objectives of this investigation is to test this hypothesis, which will be discussed in detail in the coming AMS meeting.

[1] Kang and Fu, Environ. Res. Lett., 2014, in review

[2] Cheng, Atmos. Environ., 2014, 92(August): 398-410. http://dx.doi.org/10.1016/j.atmosenv.2014.04.031

[3] Cheng and Lin, J. Geophys. Res. - Atmos., 2001, 106(D19): 22,871-22,886.

[4] Lin, Cheng, and Schroeder, Atmos. Environ., 2001, 35(6): 1141-1154.

[5] Draxler, 2013, http://www.arl.noaa.gov/documents/reports/hysplit_user_guide.pdf paper on HYSPLIT

[6] WRF 3.6 model, 2014, http://www2.mmm.ucar.edu/wrf/users/docs/user_guide_V3/ARWUsersGuideV3.pdf