CEFI Portal Data Q&A

[chiaweh2]

Hi All,

As a starter please feel free to put any discussion points or inputs/comments here.

[btupper]

Hi,

It's not clear (to this newcomer) what to make of the forecast data architecture. For example, what is the end-user's interpretation of "member" here. Perhaps an end-users field guide or glossary might be helpful?

Cheers,
Ben

[chiaweh2]

Absolutely, thank you for your input! We’re currently working on incorporating this specific information into the CEFI Cookbook. Stay tuned for updates!

In the meantime, to quickly address your question: a "member" refers to an ensemble member of the forecast. Each ensemble member is created by introducing small perturbations at the initialization field, allowing the forecast to display a "spread" of possible outcomes in the future. This allows the "probabilistic forecast" instead of a "deterministic forecast"(single result). A visual example can be seem here

This is a sea surface temperature(sst) field initialized at March 2022. The map is showing the sst field at lead time = 0.5 month. The bottom left shows the spread of forecast(light purple) that I was referring to. By calculate the mean of the spread time series (dark purple) you have the "ensemble mean" forecast. Hope this is helpful! And please feel free to let me know if you have further questions

[btupper]

That makes perfect sense - thank you!

But of course it raises another question - the lead dimension is units of months which in the file are 0,1,...,11. Where does the 0.5 month shift come in? Is that just so your displays center on the month label? Or is the lead computationally shifted by 2 weeks? For instance, is lead = 0 with an intialization date of 2022-03 referring to the period March 1 through March 31?

[chiaweh2]

Yes. It is a time label decision. What you described is exactly right.