WRF-Chem Performance with Hourly and 12-Hourly Emissions

[Naser Mohammadzadeh]

Dear all,
Hello.

I have a question regarding the use of different emission time resolutions in WRF-Chem.

So far, I have been using 12-hour emissions (wrfchemi_00z_d01, wrfchemi_12z_d01) for both chemistry schemes (MOZART-GOCART and RACM-MADE-VBS). My results generally show underestimation and very low correlations for most parameters, especially PM10 and PM2.5.

Based on this, I hypothesized that using hourly emissions might improve the results. However, after switching to hourly emissions, the model performance became even worse. I am unsure whether this is because the model is not correctly reading the hourly emission files, or whether 12-hour emissions fundamentally perform better than hourly emissions in my case.

In addition, I generated emissions on an hour-by-hour basis. I have heard that “hourly emissions” may sometimes refer to emissions provided at 00, 06, 12, and 18 UTC, rather than truly hourly emissions (00, 01, 02, …, 23). Could you please clarify which approach is correct for WRF-Chem?

I would greatly appreciate your guidance on this issue.
Thank you very much.

[Gabriele Pfister]

Hi 

you will need to provide more information. What emission inventory are you using and what is your model domain and what are the most relevant sources in your region (is it anthropogenic or are there also significant impacts from biogenics,  fires or dust)? And what tool do you use to prepare your emissions? 

Also keep in mind that anthropogenic emissions are only one source of uncertainty in the model. 

Gabi 

--

================================
Gabriele Pfister

Atmospheric Chemistry Observations & Modeling (ACOM) Lab

NSF National Center for Atmospheric Research (NCAR)

Email: pfi...@ucar.edu

Work Phone: +1 303 497 2915

Web: https://staff.ucar.edu/users/pfister

[Naser Mohammadzadeh]

Dear Pfister,

Thank you for your reply, and I apologize for my earlier incomplete question. Please allow me to clarify my experimental setup and objectives.

I am using WRF-Chem with the MOZCART option (MOZART_GOCART; chem_opt = 112). Initially, I used the EDGAR-HTAP 2010 anthropogenic emission inventory provided through the official WRF-Chem resources. These emissions were applied as 12-hourly inputs (wrfchemi_00z_d01 and wrfchemi_12z_d01).

More recently, a research group at my university developed a local emission inventory, which I am now using for comparison.

My model configuration includes two domains:

• Domain 1 (d01): East Asia

• Domain 2 (d02): South Korea

I will attach figures showing the domain configuration in this email.

Regarding other emission sources:

• Biogenic emissions: included using MEGAN v2.04

• Fire emissions: not used

• Dust emissions: turned off (dust_opt = 0)

For the local emission inventory, the specifications are as follows:

• Inventory name: UNIMIX v3.5.1

• Data periods: January–March 2022 and January–March 2024 (including 7-day spin-up), for both d01 and d02

• Spatial resolution: 27 km (d01), 9 km (d02)

• Temporal resolution: 1-hourly

• Chemical mechanism: SAPRC99 AERO4

The main objective of my study is to evaluate how different emission inventories affect model performance, specifically comparing EDGAR-HTAP 2010 with the local emission inventory, and to assess potential improvements in simulated air quality.

In addition, my simulations focus on the AsiaAQ-2024 period (15 February–13 March). I am also investigating the impact of emission temporal resolution, particularly:

1. The difference between using 12-hourly versus hourly emissions

2. Whether “hourly emissions” strictly represent hour-by-hour variability, or if they are derived from coarser temporal profiles (e.g., 6-hourly or 12-hourly distributions)

I would greatly appreciate your advice on these points, especially regarding best practices for emission temporal resolution and its contribution to overall model uncertainty.

Thank you very much for your time and guidance.

Best regards,
Naser

[Gabriele Pfister]

Hello

I don't think we can help you with your question as this is not a technical issue but more a research question and requires detailed analysis of your model output. You will also have to conduct a comparison between EDGAR and your local emissions as the difference between these two is more than just in the diurnal cycle. 

Gabi