Day 2 - HFIP Annual Sessions 3 and 4:
Session 3 Summary & Key Ideas:
Two versions of HAFS (A and B) transitioned to operations in June 2023. Three-year (2020-2022) retrospective evaluation demonstrated that HAFSv1 improved forecasts compared to legacy operational model HWRF. In 2023 hurricane season, HAFSv1 has further proved superior track and intensity forecast skills over HWRF, improved track forecast skills by more than 15% after day-2, and intensity forecast skills by ~10-20% after day-3 for storms in NHC AOR.
A number of HReX experiments were also supported by HFIP in 2023: HAFSv1.1A, HAFSV1.1B, HAFSv1.1D, MOM6-HAFS, GFS-GSL, and GFDL T-SHiELD. These experiments serve as a testing and proving ground for novel improvements to PLB, cumulus and microphysics parameterizations, advances in DA, and ocean coupling, amongst others, which are being considered for 2024 implementations. The HAFS Ensemble on the Cloud (HERC) experiment was also highly regarded by NHC, and much positive feedback was received with respect to the new probabilistic forecast guidance debuted as a part of HERC.
Session 3A: HAFS Implementation 23/24
Session 3B: HFIP R2O & O2R
Session 4A: Hurricane Observations Panel
Session 4B: DA and Impacts
Session 4C: Focused DA Discussion & Lessons Learned
- JEDI Status @EMC
- Key Points
- Lessons Learned, takes lot of time, hands on, get to Agile Mentality
- Initial HAFS DA Priorities
- Discussion
- Frank, Been piecemeal too long
- What is Best Solution moving forward for DA.
- Suspend Physics a yr?
- Gopal, Don't jeopardize forecast improvements as physics is low hanging fruit.
- Don't have to replicate, can set goal to whatever we want. Don't have to do what have now. Priority is Key
- Frank: Need to build DA Capacity Not integrated w/ anything. HPC able to do both? Probably Not. Research Community Help? Both on a shoestring or pick & chase.
- Simpler Model System would make this easier -John.
- Frank: Do we have Internal Capacity? Must Decide.
- Gopal: How do we link all these resources?
- With Supplemental ending more constrained. If individual research is possible, great but what about Integration?
- Joint Center Representative have? Have to own it ourselves. Requirements / Needs must be written clear & then presented.
- GFS Transition to JEDI helps. Need a similar plan for HAFS. Are there pieces we can leverage?
- How do we determine what resources before need/priorities?
- Can't decide in this room alone. True but we can define requirements!!!
- What effort, time, resources -> Transition Plan then take across NOAA
- Will need help & other Avenues
- Online Comments - Help from other groups to fund for us? Yes. (JTTI, WPO, etc.)
- Won't fund Transition if JEDI isn't ready vet. Funds can transition but not finish the build.
- Academics are funded though. No Support to this today
- Dev, Limited Community, ask HPC to bring in players w/ Training? Frank will work on workforce development
- DA Consortium? Who leads?
- Applications in. Challenge is HPC.
- Summary requirements Internal & Ext
- 1 Grant over Several Years.
- For storms under RI, comfortable where at? In walk away from HWRF.
- Balance JEDI/HAFS? Pause? When?
- NHC-Take advantage of non-Gaussian Methods. GSI vs JEDI? When Pivot?
- Matt (NHC) - Consistency good for Forecasters. Willing to wait.
- Zach (EMC). Not Satisfied with current performance
- Take a parallel track. Community must help us.
- How many FTEs will it take? Don't know, don't sell short.
- Change cadence to 2 years from 1?
- Develop transition plan & req - Frank
- Can we do live w/ DA as is while all on JEDI DA Development?
- Do we hold HAFSv2 for now for JEDI? -> Some pieces are ready now and only need testing. Then stop.
- Initial Jumpstart needed on JEDI, can't depend on students, need internal staff able to code/learn.
- John (UMD) - Faster better is painful without bringing forward ie GSI to JEDI. Morale hit.
- Shu Yan - If JEDI is part of UFS, what support can we expect? Community level support?
- Yes is agreed, the community is complicated.
- Get a copy of the current transition plan? Yes, Recommend a Tiger Team Quick.
- Arette is on it, Frank