Using Tempest Extreme to track (Anti)Cyclones anywhere on any level surfaces!

Some takeways from the 2025 Princeton GFDL Hackathon, esp. on tempest extreme, a tracking algo.

The amazing group of hk25-blocking worked jointly on identifying blocking (persistent mid/high latitude anticyclones) and (extra)tropical cyclones. Members include Dr. Gan Zhang from UIUC, Dr. Mingyu Park from Princeton, Yanjun Hu from Purdue, Coleman Gliddon from MIT, and me.

A tool to track any features that depart from neighboring points (say, cyclones on surface pressure, or anticyclones on z500 isobaric level, etc.) could be using this tool of Tempest Extreme, results can be downloaded from here:

1. xsh24 surface cyclone tracking (200 Pa increase outward within 6 degs) [GFDL downloadable from /scratch/cimes/xy4043/tracks]
2. xsh24 z500 anticyclone tracking (10 m z500 decrease outward within 6 degs) [GFDL downloadable from /scratch/cimes/xy4043/tracks]
3. ERA5 z500 anticyclone tracking (1990-2020) (10 m z500 decrease outward within 6 degs) link
4. Filtered Euro-Atlantic ERA5 z500 anticyclone tracking (1990-2020) link

…and here’s a quick tutorial on how to use the tool:

(Note: Performance comparison assessments of results compared to blocking tracking algo ConTrack is also done by Mingyu.)

The following tutorial uses stellar-amd node of GFDL and runs on the X-Shield 2024 data that is not published yet.

1. module load anaconda3/2024.10
2. Install the yml file via: conda env create -f env_tempest.yml
3. Register as an ipynb kernel: conda activate tempest_extreme_new python -m ipykernel install –user
   –name tempest_extreme_new
   –display-name “Python (tempest_extreme_new)”
4. Check nco
5. Check cdo
6. Go to /scratch/cimes/xy4043/tutorial and get all .sh files (or visit the hackathon blocking group repo)
7. chmod +x stitch_nodes_3hrly_anticyclone.sh (and all other 3 sh files)
8. run ./detect_node_anticyclone.sh
9. run ./stitch_nodes_3hrly_anticyclone.sh
10. The results (raw, and reformatted csv) are in /scratch/cimes/xy4043/tracks/
11. If you want to extract specific tracks that pass through some lon lat boxes, run filter_tracks.py (email Ken for code)

If we impose certain spatialtemporal constraints, we can filter out persistent blocking events:

The xsh24 model also has several temperature tendecy terms, and here’s a quick note:

1. shallow convection is called ‘tendency_of_air_temperature_due_to_shallow_convection_coarse_C3072_1440x720’
2. deep convection is called ‘t_dt_gfdlmp_coarse_C3072_1440x720’, mp = micro-physics
3. temperature tendecy ‘t_dt_phys_coarse_C3072_1440x720’ includes both radiation and shallow convection
4. More to search up from tendency_of_air_temperature_due_to….!