Back Trajectory Source Region

[jdreessen]

The project I am working on utilizes back trajectories to analyze pollution sources contributing to ozone in my area. Overall the goal of my group is to identify major players (e.g., power plants, industry, etc.). I am hoping to give reliable geographical data to guide my group's investigations and identify large NOx emitters contributing to the following day's ozone. Please see the image I have attached. The blue area is, as of now, the area I am recommending they limit their investigation on this particular day. However, I would like to reinforce this methodology with more rigorous scientific support of the Hysplit Model.

I have run several 24 hour back trajectories during the course of the day in question (e.g., July 2, 06, 09, 12, 15, 18, and 21UTC all at 1000m AGL). This gives a spread to the trajectories and provides a larger area from which to consider emissions. Reading through other forums and FAQs I have gathered that I should also do additional trajectory levels to better approximate this spread. I have also read that trajectory error may be as much as 15-30% of the travel distance.

What I have done in the image attached is create a buffer around each trajectory then merged those areas in to one shape using GIS software. This shape will be given to compliance/policy colleagues who will use it to determine if large NOx emitters fall within this area. Currently I have a buffer (half width) of 15km at night expanding to a buffer of 30km during the day to try and account for additional mixing. My concern is this is methodology and distance is quite arbitrary. Is there a more standardized method to assess the "area of influence" around a trajectory?

[barbara.stunder]

Using trajectories to estimate source regions is limited because the trajectories do not account for turbulence. Certainly using different levels in the boundary-layer would help. Also the buffer should increase with distance along the trajectory because of the error of 15-30% of travel distance. A different approach, if you know the locations of the large emitters, is to run forward dispersion or trajectories from them.

[jdreessen]

Thank you for your reply barbara.stunder. The forward trajectories from large emitters is an excellent idea. Unfortunately, my group is relying on the back trajectory methodology to identify emissions sources and these areas may include upwards of 100+ different sources. I was hoping for information in the way Hysplit calculates trajectories, either procedurally or scientifically, or previous work from others that strengthens the justification for the areas I identify.

Your suggestion to run forward trajectories from significant point sources as a confirmation, once they are identified, is probably the ideal way to confirm the back trajectories. But at what distance do I consider the forward trajectory as "relevant" to my area. Seems there would be the same "area of influence" question as with back trajectories. Dispersion modeling from each point source would help solve this, but if I can justify sources without running several dozen dispersions, it would be advantageous.

Thanks for your feedback.

[barbara.stunder]

For information on the trajectory calculation, errors/uncertainties, and source attribution see the workshop training information at http://www.arl.noaa.gov/HYSPLIT_workshop.php.

If you ran forward dispersion knowing the source term (amount of pollutant) then you would know the concentration at your receptor. In your case there is chemistry involved, resulting in ozone, so it is not a simple modeling exercise.