make --closedcontourcmd resolution specific

[chengzhuzhang]

Description

TC detection threshold should be model data resolution depended, currently the pre-processing script is using a threshold for ne120, resulted small number of detected TC events in ne30 simulations. Now update to add resolution specific thresholds. Note: this fix should bring into cdat migration branch, and zppy implementation.

The original closed contour criteria (--closedcontourcmd "PSL,300.0,4.0,0;_AVG(T200,T500),-0.6,4,0.30"): requiring an increase in PSL of at least 300 Pa (3 hPa) within 4 Degree of the candidate node, and a decrease in 200 and 500 hPa average air temperature of 0.6 K within 4 Degree of the node within 0.3 degree of the candidate with maximum air temperature.

Checklist

• [ ] My code follows the style guidelines of this project
• [ ] I have performed a self-review of my own code
• [ ] My changes generate no new warnings
• [ ] Any dependent changes have been merged and published in downstream modules

If applicable:

• [ ] New and existing unit tests pass with my changes (locally and CI/CD build)
• [ ] I have added tests that prove my fix is effective or that my feature works
• [ ] I have commented my code, particularly in hard-to-understand areas
• [ ] I have made corresponding changes to the documentation
• [ ] I have noted that this is a breaking change for a major release (fix or feature that would cause existing functionality to not work as expected)

[chengzhuzhang]

With --closedcontourcmd "PSL,300.0,4.0,0;_AVG(T200,T500),-0.6,4,0.30" change into --closedcontourcmd "PSL,300.0,4.0,0;_AVG(T200,T500),-0.6,4,1.0", both threshold somehow created same results.

Changing --closedcontourcmd "PSL,300.0,4.0,0;_AVG(T200,T500),-0.6,4,1.0" to --closedcontourcmd "PSL,300.0,4.0,0;_AVG(T200,T500),-0.6,4,1.2", slightly modified the distribution.

[chengzhuzhang]

Below is a comparison between v3.LR and the base line simulation (v2ish) With threshold 1.0 only 5 events are detected in 15 years of data LR v3.historical (2000-2014). Results are similar with threshold 0.3.

With the investigation so far, it is determined that:

• At LR, the detection is largely insensitive to the threshold change.
• With same threshold, v3.LR detected much less storms than in v2ish simulation with this algorithm.
• We should test when v3.HR data made available.

[chengzhuzhang]

@forsyth2 Just a heads-up that I updated the pre-processing script for TC detection, so that the detection is model resolution specific. Also the bug that the mesh generation file can't be written to cfs directories should also be fixed. We should plan to update zppy accordingly with these changes.

[forsyth2]

@chengzhuzhang How would you like me to review this PR?

• Visual inspection?
• Testing this myself?
• Implementing corresponding changes in zppy and testing those?

[chengzhuzhang]

Visual inspection would be nice. Mostly, it is a heads-up that we should propagate this change to zppy, and hopefully can resolve some recent issues with the tc_analysis in zppy.

[chengzhuzhang]

Looks good from visual inspection thanks. I'll make/test the corresponding changes in zppy too (which might even help with E3SM-Project/zppy#180 & E3SM-Project/zppy#622)

Thanks! @forsyth2