Methods for initializing isotopes within the Ice Sheet

[Katetc]

After a conversation with @whlipscomb yesterday, we have many thoughts and questions surrounding the appropriate way to initialize the isotopes throughout the ice sheet. This issue could be a good place to discuss the needed science and best ideas for initializing the isotopes throughout the icesheet. Here are a couple of thoughts we had recently.

1. It is certainly possible to describe a high resolution 3D field that can be read in from a file to initialize the ice sheet. This is most likely the restart or initialize from a restart option in most cases. We could create the file or create a tool for creating a file like this from "data" of some sort as well.

2. Currently, there is an "ice age" tracer field in the ice sheet, and it is set to zero at start up. Old ice age is updated every year and new ice is initialized at "zero" age. We have some spun-up ice sheets with age or initialized age files and could create a table of some sort that associates isotope concentrations with ice age. However, the age of ice is kind of logarithmic, so the table refinement per year may change in certain ranges. For instance, we might have more isotope concentrations associated with ages 0-1,000 years than 1,000-2,000 years, and even fewer for 20,000-100,000 and fewer still for 100,000+ year old ice (very old ice).

3. The ice sheet could be initialized without tracers and run with constant climate forcing for approximately 20,000 years to filter isotopic concentrations throughout the ice sheet. This may not be enough to reach all areas, but should be pretty good. We would probably need time or climate varying isotope forcing with the SMB at the surface so the ice sheet is spun up with something resembling the correct isotopes for the climate of the time.

Other ideas? @whlipscomb @billsacks others?

[billsacks]

Thank you both for laying out these thoughts!

Another simple method that I think we might apply in other places is to initialize isotopes with constant ratios equal to their average global ratios, and then spin up to get a better distribution of isotopic concentrations. (Actually, for numerical reasons, we will probably use a ratio of 1 for every isotope, and then these will be converted to scientifically-meaningful isotopic ratios in post-processing.) However, I realize that there could be extra complexities for the ice sheet because of its long spin-up times.

Also including @jiang-zhu.

[jiang-zhu]

Thank you, @Katetc, @billsacks, @whlipscomb !

The offline spin-up approach sounds reasonable to me. Generating the required isotope boundary fields should be straightforward using iCAM5/iCAM6 simulations nudged to ERA5.

We do have ice-core records going deep into the ice sheets, which could either be used to initialize the model or as an independent benchmark for validation and evaluation.