Confusion about Model Operation and Atmospheric Forcing Data

[yuezhouchen]

Hi all,

I am writing to seek your insights regarding some uncertainties that I have encountered while running the model. My concerns mainly revolve around the integrity of forced atmospheric data.

Let's assume that my simulation time frame is set from 1980 through 2020. In this case, would I need to input complete meteorological data for these forty years? How does the model handle any missing data, say, if a year's or even a month's data is unavailable or missing?

Another question I have is about the model's predictive capabilities. Is it possible to use it for short-term future predictions? For instance, if I desire to simulate data from 2010 to 2025, but lack meteorological data for the future (beyond July 2023), could the model still operate effectively? What level of credibility can we assign to such results?

These are some thoughts that suddenly sprang into my mind and I hope someone could help clarify them. I deeply appreciate your cooperation and participation in discussing these issues.

Best Regards,

chen

 

[islas]

Hi Chen,

You do not provide CESM with meteorological data. It is a physics based model that generates its own weather internally. The inputs to the model are external forcings such as time varying anthropogenic and natural changes in atmospheric constituents (CO2 and other greenhouse gases, aerosols etc) and time varying insolation and land-use and land-cover change. For future simulations, projection scenarios are used to specify the forcing and the model, again, generates it's own meteorology. Because the model generates its own weather, you do not expect any given day or year to look like what happened exactly in observations, but with an ensemble of simulations the aim is that the statistics of the climate system would resemble that of the real world. There are already simulations available as you describe that you can analyze e.g., CESM2 Large Ensemble Community Project (LENS2) | Community Earth System Model.

It's possible that I have misunderstood the question and you're actually wanting to run CESM with specified dynamics or nudging in the atmosphere. If that's the case, then you would need the observed meteorological fields to nudge towards and you couldn't do that in the future. But since you didn't mention that you're wanting to used specified dynamics, I'm assuming that's not your question.

Alternatively, maybe you're wanting to run offline land model simulations, in which case you would also need the meteorological forcing and I don't think it could handle missing data, but I'm not certain about that.

Isla

 

[yuezhouchen]

Hi Chen,

You do not provide CESM with meteorological data. It is a physics based model that generates its own weather internally. The inputs to the model are external forcings such as time varying anthropogenic and natural changes in atmospheric constituents (CO2 and other greenhouse gases, aerosols etc) and time varying insolation and land-use and land-cover change. For future simulations, projection scenarios are used to specify the forcing and the model, again, generates it's own meteorology. Because the model generates its own weather, you do not expect any given day or year to look like what happened exactly in observations, but with an ensemble of simulations the aim is that the statistics of the climate system would resemble that of the real world. There are already simulations available as you describe that you can analyze e.g., CESM2 Large Ensemble Community Project (LENS2) | Community Earth System Model.

It's possible that I have misunderstood the question and you're actually wanting to run CESM with specified dynamics or nudging in the atmosphere. If that's the case, then you would need the observed meteorological fields to nudge towards and you couldn't do that in the future. But since you didn't mention that you're wanting to used specified dynamics, I'm assuming that's not your question.

Alternatively, maybe you're wanting to run offline land model simulations, in which case you would also need the meteorological forcing and I don't think it could handle missing data, but I'm not certain about that.

Isla

Thank you for your generous response. I had previously speculated that the model would read the meteorological forcing data I input, and then make adjustments to its weather simulations based on this. If my understanding is correct, your answer seems to confirm this. You're right in guessing that I am running CLM offline. I do have another question: The amount of atmospheric forcing data accepted by the model doesn't really have a strong connection to the time period being simulated by the model, as long as they're within the same timeframe. For instance, if I were to set up a simulation from 2020 to 2025 and only provide data up until 2023, the model could still fully operate autonomously, as it has its own meteorological system. The user-inputted data is essentially just there to correct it and make it more aligned with reality. I presume this understanding should be correct.

 

[islas]

If you're running CLM offline then that is a different story and you would need forcing data for the full time period. Offline CLM doesn't generate its own weather. You need to be fully coupled to the atmospheric model for that. So you would need to give the model something from 2023-2025.

 

[yuezhouchen]

If you're running CLM offline then that is a different story and you would need forcing data for the full time period. Offline CLM doesn't generate its own weather. You need to be fully coupled to the atmospheric model for that. So you would need to give the model something from 2023-2025.

So that's how it is, I understand now, thank you very much.