Merge pull request #2 from nsidc/add-notes-20230914

Add notes from Thursday's technical meeting

doc/_notes/20230914-technical-meeting.md

@@ -0,0 +1,97 @@
+# Notes
+
+## 2023-09-14 meeting with Liz, Mike, Walt
+
+* The README in this repo is out of date. Liz has a newer one, will send. See
+  e-mail!
+* Code runs in steps. Liz wonders if `update_data.py` is meant to run them all in one.
+  Based on notes below from Mike, maybe not?
+	1. Download all NSIDC-0080 faily granules using `nsidc_download_Tb_data.py`
+		* For data before 2016, `NSIDC-0001` and `NSIDC-0007` are used.
+	2. Generate all the daily melt binary files using `generate_daily_melt_file.py`
+		* Binaries provided in `data/daily_melt_bin_files` already calibrated by Tom
+			Mote for pre 2016 dates.
+	3. Run the main script (`main.py`) with `--preprocess` flag. Creates
+		 climatologies, other things. May not need to be done every day.
+	4. Run the main script (`main.py`)
+* Liz emailed a 2022 threshold file, need 2023 from Tom Mote for forward
+	processing. Ted or Walt will reach out to Tom for this file.
+* Liz thinks we can skip the preprocess step for these daily images. In fact,
+  may only need to run `generate_antarctica_today_map.py`. She thinks line plots
+  (`plot_current_year_melt_over_baseline_stats`) and maps can be generated with this
+  code!
+	* Mike: Generating daily files builds 4 things:
+		* daily melt map
+		* cumulative melt map (from Oct 1 to current day)
+		* melt anomaly map (req. climatology)
+		* line plot (req. climatology)
+  * Mike: Climatology is initialized, which can take a long time, but after that it can
+    be incrementally updated.
+* Many parameters have defaults that Liz would like to extract to constants,
+	e.g. `doy_start` and `doy_end`.
+* Ted would often ask to have both full-extent and peninsula-only maps, would
+	be nice if that was parameterized (pick a region-number; what about manual
+	bbox?). Mike has already done this!
+* One of the data files contains region definitions.
+* At runtime, a giant pickle file is generating containing the full historical dataset
+  (3d array)
+  * Downloads full dataset from NSIDC.
+  * `update_data.py` will update the pickle file every day with new data.
+  * Other intermediate CSVs are generated.
+
+
+### Liz experiences
+
+* Liz never created the figures I'll need to create. The code I need to run
+  should be in this repository. Where?
+* Liz created a document for me, will send. See e-mail! They may be cryptic,
+	ask her if I need more info. This document is about the plots _she_ created,
+  which will be different than the ones I create.
+
+
+### Mike background
+
+* Wanted to automate things initially.
+* Annual thresholds generated by model runs by Tom Mote. Code is on his machine, but
+  nobody else has seen it. Close to retirement, no hand-off plan.
+* He may be open to opening his models/code.
+* Mostly IDL, Fortran, shell scripts. Built in 90s, has undergone many changes over
+  decades.
+* Tom may apply some personal expertise to subjectively adjust values.
+
+
+### Scope of the work
+
+Create daily images akin to Greenland Today. Runnable by ops.
+
+* Maps
+	* Cumulative melt: How many days experienced melt for each grid cell?
+	* Daily melt: Which grid cells melted this day?
+* Plot
+	* Line plot: Total extent (`km^2`) of melt on a given day. Maybe % of total?
+		Greenland Today uses `% of total ice sheet`, but this may not be suitable
+		for Antarctica [Walt]. Maybe make it easy to do both.
+
+
+### Questions
+
+* Where to write the images?
+* 
+
+
+## Repo plan
+
+* Migrate to NSIDC
+* Notify Mike for PRs (set up in GitHub)
+* Document Mike as co-maintainer with NSIDC
+* Loop Mike in for questions, could help with unit test cases
+
+
+## TODO
+
+* Email Mike:
+    * Shared repo for AT code
+    * Mike will be admin, maintainer
+    * Instructions for transferring repo
+    * Instructions for GitHub notifications
+* Mike uses Slack, consider invitation: [email protected]