Hello,
I am a researcher studying air quality in South Korea. I have been exploring the MPAS-CMAQ model and have successfully completed the example simulation following the user guide.
Now, I am attempting to run the model for my specific research period. My goal is to evaluate the MPAS-CMAQ simulation performance for January 2017. I have downloaded all the necessary input data, including emissions, restart files, and FDDA files based on the provided resources.
However, I have a few questions regarding the MPAS-CMAQ process:
Currently, it appears that the provided FDDA files only cover the u, v, T, and q variables.
(1) Generating FDDA files for additional variables:
I would like to perform FDDA for variables other than u, v, T, and q. Could you please guide me on the procedure to achieve this? Would this require using the latest MPAS version (e.g., v8.x) or specific pre-processing tools?
(2) Use of Global Reanalysis Data (FNL/ERA5):
In WRF-CMAQ, it is common to download global reanalysis data (such as FNL or ERA5) to drive the model via nudging. Is a similar workflow required or recommended for running MPAS-CMAQ? I would like to clarify if I need to generate my own nudging fields from global models for my specific period.
Any insights or advice on these points would be greatly appreciated.
Thank you.
Hi Jeong Beom,
We have added FDDA feature in WRF many years ago by one of my colleagues, Russ Bullock, who retired last year. Later, he implemented the same thing in MPAS. The implementation was based on u, v, T, and q. If you want to add a new variable, please examine his implementation in MPAS.
For your second question, yes you need to create your own FDDA files. Our files were created by another colleague, Jerry Herwehe, who retired last year as well, with some ad hoc scripts. Unfortunately, we won’t be able to share those scripts.
If you have additional questions, please let us know.
Cheers, David
Thank you for answering my questions.
Your response has cleared up the confusion we had on our end regarding the FDDA files.
Although I have never generated FDDA files myself, I see this as a great opportunity to learn and challenge myself. I believe tackling this will be a valuable experience.
I may have further questions after I look into the FDDA generation process and emission species mapping on my own. For now, I do not have any additional inquiries.
I would like to thank the CMAS Center for providing this platform for communication. I also want to extend my sincere gratitude to you, Jerry Herwehe, Russ Bullock, and everyone involved in the development of the MPAS-CMAQ model.
Sincerely, Jeong-beom Lee
1 Like
Dear David,
Following your previous advice, I am currently working on global MPAS-CMAQ simulations and specifically studying the implementation of FDDA.
As I understand it, FDDA in numerical models like MPAS or WRF works by applying a weighting factor to global meteorological variables (u, v, t, q) to nudge the simulation toward observed or reanalysis data. However, while preparing for global-scale runs, I am concerned that continuous FDDA might cause the MPAS output to become overly similar to the driving global meteorological data, potentially diminishing the model’s own predictive role.
I would like to ask the following questions:
1. Similarity between MPAS results and global analysis (FNL) data
The MPAS-CMAQ guide provides FDDA data at 6-hour intervals for an entire year. If I use this setup, does the MPAS meteorological output essentially converge to match the global analysis data (such as FNL)? I am curious to what extent the model maintains its unique simulated characteristics under this configuration.
2. Model stability without FDDA
If I perform an MPAS simulation using only a hot-start initial condition and the model’s internal physical mechanisms—without enabling FDDA options—is it likely that the results will diverge or produce unrealistic meteorological fields over a long-term global run?
If such divergence is a significant risk, I assume applying FDDA is necessary. In that case, what would be the most appropriate time interval for FDDA input to balance model stability with physical autonomy?
I am still learning a lot, so I truly appreciate your patience and your expertise in answering these questions.
Dear Jeong Beom,
I understand your concern. Indeed, you can control the strength of nudging (provided/included in the released run script). Here is the answer for your specific questions.
1. Similarity between MPAS results and global analysis (FNL) data
It won’t since a very weak nudging values are used.
2 Model stability without FDDA
We run our model always with FDDA files so I don’t know what will happen without them. But when I worked on the WRF-CMAQ coupled model, I accidentally set the nudging for only the first 4 days, and we ran a simulation for about a month. We saw significant divergence after about the 10th day when comparing with observations.
In a nutshell, we apply very weak nudging and still let the dynamics perform its job.
Cheers, David
1 Like
Dear David,
Thank you for your prompt response. I now clearly understand the points you explained, and I have no further questions regarding FDDA at this time.
Lastly, I would like to express my sincere gratitude for the paper you and your team authored, “Development of the MPAS-CMAQ coupled system (V1.0) for multiscale global air quality modeling.” It has been incredibly helpful for my research.
Thank you again for your time and assistance.
Best regards, Jeong-Beom Lee
1 Like