Expected Transport Distances for NOx, SOx, and PM2.5 in InMAP

[Zoe Wang]

Hi everyone,

I’m using InMAP to model emissions from a coal plant on the Germany–Poland border. With a threshold of 0.001 µg/m³, I’m seeing:

• SOx (7,500 tons) dispersing ~200 km

• NOx (11,500 tons) reaching ~1,500 km (as far as Greece)

• Total PM2.5 exceeding 1,000 km

Do these transport distances seem reasonable? Or could I have misconfigured something? Thanks for any insights!

Zoe Wang

[Chris Tessum]

Hi Zoe, 

Sorry for the delayed response. I guess those transport distances don't really align with my expectations of what I would guess they would be. Keep in mind that InMAP is a reduced complexity model, and its performance has not been carefully evaluated in Europe (at least not by me). If it's not giving results with sufficient mechanistic realism for your use case, that may mean that you need to use a more comprehensive model. If you're interested in a modeling system that can theoretical provide more realism while still being easy to use and reason about, we have a project about that here: https://earthsci.dev/dev/, although it is still very much in an in-development status. (It doesn't currently do what you want, but we're working in that direction.)

Best regards,

Chris

[Zoe Wang]

Hi Chris,

Thank you for pointing me to EarthSciML. This project looks very relevant to my research, and I’d love to stay updated on its progress.

Regarding InMAP, I also tested the model on a toy example of U.S. emissions using different atmospheric chemistry model simulation data as input (GEOS-Chem and WRF-Chem), while keeping the configuration file and emission data the same. These input datasets were collected from the InMAP package releases page.

The figure below presents the results from the two simulations. From this comparison, I noticed:

1. Simulation differences mainly stem from the chosen atmospheric chemistry model input.

2. Results differ greatly in SOx dispersion distance and concentration.

3. Total PM2.5 and NOx results are relatively consistent across both input models.

Thanks again for your insights!

Best,
Zoe