Committee members attending: Chuck Pavloski (SAC Chair, outgoing), Victor Gensini (SAC Chair, incoming), Kevin Goebbert (Usercom Chair), Ryan Abernathey, John Allen, Dave Blodgett, Eric Bruning, Sen Chiao, Casey Davenport, Alex Davies, Redina Finch, Charles Graves, Alex Haberlie, Scott Jacobs, Aaron Kennedy, Todd Murphy, Anne Myckow, Jim Potemra, Craig Ramseyer, Justin Rice, Keah Schuenemann, Kim Wood
UPC staff attending:
Mohan Ramamurthy (UPC Director), Ryan May (UPC Deputy Director), Drew Camron, Shay Carter, Julien Chastang, Stonie Cooper, Nicole Corbin, Ethan Davis, Doug Dirks, Tara Drwenski, Steve Emmerson, Ana Espinoza, Ward Fisher, Dennis Heimbigner, Yuan Ho, Hailey Johnson, Megan Lerman, Thomas Martin, Tiffany Meyer, Joshua Neigigh, Jennifer Oxelson, Matt Perna, Inken Purvis, Sheri Ruscetta, Mike Schmidt, Tanya Vance, Jeff Weber, Tom Yoksas, Mike Zuranski
Introductions around the room
Thanks to parting/transitioning members and welcome to new committee members (Mohan Ramamurthy)
Update by committee chairs (Kevin Goebbert and Chuck Pavloski)
View Dave’s presentation.
View Anne’s presentation.
View Justin’s presentation.
View Bernard’s presentation.
View Scott’s presentation.
View Mohan’s presentation.
Unidata is in the process of creating a new strategic plan, which will inform the next core funding proposal to the National Science Foundation, to be submitted in late 2023.
View Tanya’s presentation.
As part of Unidata’s efforts to be more effective in delivering services to the community, we are trying to define what activities are included in the scope of those services. The committee took part in several exercises to illustrate some of the issues UPC staff members face when interacting with community members, and gave feedback on the situations described. This input will feed into a document under development that will provide both staff and community members with guidance about what to expect.
ChuckP: Off-site indirect rate? Is that new?
JimP: off-site employees pay their own utilities. Can employees get reimbursed?
Joshua: UCAR provides stipend to employees for home offices
JimP: University of HI has indirect cost rate 23% for off-site employees
MohanR: Unidata has only recently started using UCAR's remote indirect rate
ChuckP: What about employee travel?
MohanR: UCAR may provide some funds, but so far those funds have come from Unidata
KevinG: What are the prospects of getting more core funding?
MohanR: We are having these conversations with Bernard Grant all the time. But it's for NSF to decide what the level of funding will be. NSF has been generous over the years, but they also have finite resources.
KevinG: Is there a target for your carryover amount?
MohanR: During government shutdows, NSF made the decision to continue funding facilities even when individual researchers are embargoed.
JimP: Maybe a strategy for the proposal -- can we capture how many proposals to NSF are citing use of netCDF or THREDDS in their data management plans? Can we mine that data and present it to the review committee?
MohanR: We don't have access to the proposals, only the summary abstracts are shared.
DaveB: I noted that you had a number of unfunded collaborations. Companies can propose against testbed/pilot funding from OGC.
EthanD: I monitor the requests, but we have not proposed.
AaronK: Many students have gpus in their gaming computers
KevinG: online asynch tutorials/materials for a variety of audiences: Folks with no knowledge of AI/ML, Folks with existing skills. Maybe also emphasize underutilized features.
VictorG: Big thanks for the awips blog, helpful for students. Is there a specific timing for version 20 for the mac?
ShayC: We are working on v20 , going from python 2 to python 3 and java 1.8 to java 11. We’re hoping for a beta "soon", will be built for mac/linux (windows may be later) hoping to release before end of calendar year.
KimW: Flipped classroom -- want more tutorials in YouTube form, students gravitate toward visual references, can they be targeted more generally? (What is IDV, What is AWIPS in general.) Personally I've shifted toward flipped classroom so in person is more interactive
RedinaF: Videos become outdated quickly. I defined an independent study around the metpy videos but they're out of date. Instructors must provide guidance on what to change.
Group: expiration dates are good.
JimP: Institution wants visual resources to be accessible (cc, etc.) so they frown on YouTube. My students: maybe 1/3 come to the classroom for lecture, some live online, the rest watch later. University likes this because it encourages nontraditional students. (But how to set tuition rates)
NicoleC: updating videos is very time consuming -- can we expand our idea about what online learning can be. E-learning modules are easier to update for software changes.
KevinG: getting students to read anything is difficult! (More so than pre-pandemic)
NicoleC: if elearning is created with instructional design principles its easier to get around
KevinG: Reach out to community members to write something "soon" rather than setting up a schedule. Reach out to community members directly.
JohnA: Generally I'm allocating time "next month" -- maybe have a range to time scales. Next month, maybe, next week? Hard deadlines
TanyaV: Maybe a rotating schedule where folks can fill in when they miht light contribute something.
DaveB: Missing piece for our community is shared testing data, especially for cloud. It would be helpful for unidata to lay the groundwork for that, making test data sets available.
HaileyJ: both netCDF teams are working on that.
KevinG: working on slightly modified language to encourage submissions, AlexD offered to help craft. What should a focus be for the next cycle?
ShayC: instruments or sensors?
KevinG: it depends, we're trying to partially broaden but specifying that unidta technologies must be included. Giving better guidance to review panels about how to judge proposals. Open but with boundaries.
KevinG: Himawari data would be nice but would be fantastic for our community
Standard visible channel, water vapoor, clean IR, want to make true color imagery.
Getting it all would not be feasible, but those would be great
VictorG: I have questions about why NOAA does what they do, specifically model data distribution. How much do we as a community need to try to influence what ocmes over conduit or nomads. GFS ensemble as an example: limited set of data coming over conduit, if you're a power user you have to go to nomads, there are redundant files hard to work with. As files get larger due to smaller grids etc., there are rate limitations on the noaa firewall. We could play a role as a community to ask why things are done the way they are.
ScottJ: I have not answer, but Anne said she's open for input on what's on conduit and nomads.
KevinG: MikeZ planning to delve into this, we have a role (esp. usercom) to shape
VictorG: nomads catalog metadata are nearly nonexistent. Documenting what should be there vs what is there is a headache. Can Unidata help?
TomY: remind everyone that the content of conduit can't be significantly increased until spring/summer 2023.
YuanH: I've been providing classroom style training courses. Any interest in remote IDV training form committee members?
SteveE: we've conducted some virtual workshops for NOAA.
DaveB: I would love for "modern, cloud based" to not mean proprietary clouds. Parallel netcdf to object shote would be be great. If netcdf could focus on those basic architectures and not on the nuances of the corporate cloud implementation. Example: amazon's way to specify credentials is leaking into how you specify a URL to an object.
RyanM: is that using an S3 compatible api?
DaveB: Yes, lots of implementations have a different back end an present as S3
DennisH: restricted by having a library to access different stores. If its s3 we should be able to deal with it without much trouble. We originally used a non-commerical system for that. If you have a specific store you'd like us to support, put up a github issue or contact us.
RyanA: as someone who has put zarr data into many object stores, in python it's easy to implement a storage backend. The constraints on a c-language library that talks to different remote storage services. Break that out into a stand-alone library that can be used by netcdf -c , but not tightly coupled. This is available in gdal? Open a conversation with the gdal folks to see if there is a common piece of software we can collaborate on.
DaveB: it would be great to have parity between gdal and netcdf c
[ACTION: netcdf team to have a conversation with outside groups about collaboration]
KevinG: costs to managing open source contributions -- how do we make community contributions to features and maintenance more sustainable? How can we support Unidata in being the community hub? Scope of support?
AaronK: we collaborate with tribal colleges In ND, brought in to help get instrumentation installed. They are understaffed and many folks used it as a stepping stone. It's a long process to keep building relationships. Sometimes EPSCOR reps reach out to us, sometimes we have a projec and we reach out, instrumentaiton vendors have contacted us after making a sale. Sitting Bull college has a masters in enviro science, have had trouble filling faculty positions.
KeahS:At MSU we're getting better at serving
JimP: UH is also MSI, sovereign rights to indigenous data.
JeffW: for decades we've been promoting open data access. This is a slightly different issue. Much of the data being collected via TIDE will be shared, but some data are not. The point is that the tribe gets to decide what is shared, they should be the gatekeeper. It's a different approach. What we don't want to see is a federal agency place sensors and take the tribale community out of the loop.
TanyaV: echoing what jeff said, and highlighting what's involved in data governance, how data are collected and shared. There may be sacred sites and history. How it's shared and how honored.
AaronK: providing notebooks that work -- maybe more education to be found in things that don't work.
RyanM: in synchronous formats we used some exercises working through errors. Baseline skill
KimW: plotting skewTs using Uwyo interface -- sometimes fails due to traffic. Having a notebook can share in advance to see errors.
KevinG: archive data ecosystem is fragile. Uwyo may choose not to continue. DarylH may not be able to provide the access he does forever. Access is difficult for older data -- real challenges. Unidata can be a convener -- help the community help itself, bring together the agencies
KeahS: undergrad education: NWS describes some of the curriculum, AMS describes some -- there is a community out here who are not yet python experts yet, and don't have a ton of room in the curriculum. NWS/AMS requirements could use some updating, maybe some lobbying.
ShayC: some amount of coding skill is needed for many disciplines. Do other degree programs require or offer an intro to programming?
KeahS: one off, not related to our data
KimW: intro classes are in C, students tell us it's useless. We have a computer methods course for met, but many students avoid it. Injecting modules into existing courses.
JimP: do you have a list of known administrators of the TDS sites?
HaileyJ: Not everyone provides that information.
DaveB: Publish a standard operating proceedure about what you are planning to do, so Unidata does not have to support as much legacy software. Many organizations break backward compatibility. Maybe a more prograessinve stance?
JimP: I think thredds is a bit different because it is a core underlying service with a lot of infrastructure.
DaveB: the thing that breaks needs to take the responsibility for updating/maintaining their code. Shouldn't expect the dependencies to continue to support unsupported operational systems. (antipattern I see a lot)
The committee separated into several smaller “breakout” groups to discuss based on the following prompting questions: