Skip to content

QARTOD 101

Jump to bottom
Kathy Koch edited this page Mar 16, 2018 · 17 revisions

Documentation from QARTOD 101 presentation/webinar

Additional Links

Questions and Comments from the webinar

Q: Many of these things are above our level of computer expertise. Do we provide you the numbers and you institute the tests? A: Yes we institute the test. We'll talk about it.

Q: Is the test parameter spreadsheet (or an updated copy) on the github site or elsewhere?

Q: Does hourly qartod testing mean all data is delayed at least an hour for GLOS products, eg HABS portal? A: No, flags are delayed, data are available.

Q: I have a couple comments about high dissolved oxygen levels in Lake Erie stations which are being flagged at various times during blooms... who's the best contact to discuss that and other test thresholds with?

Q: How can application of QARTOD help with non-sudden QA degradation (e.g., drift due to fouling)?

Comments and additional questions:

  • Let IOOS know if you have an additional parameter / variable for QARTOD to address. Provide examples of test results to be incorporated into manual updates.
  • Hey, I'm a journalist who was invited to join this and really appreciated that. I'm not at all surprised most of the material was beyond my expertise; I was expecting that. What I'm trying to do - as time allows - is learn a little more about how scientists do their jobs. Thanks for the opportunity and I hope to do this again, when I can work webinars like these into my schedule.
  • I haven't heard much about buoy cams. At one point there was work towards camera sponsorship etc. but haven't heard updates...
  • Not all flat line is a fail. Fail to me means bad data. So that is dangerous. I have seen many cases where water temp, DO etc stay in one value for more than 4 readings. But that goes back to debates during Qartod development.
  • I am not sure, but we can discuss this later. Since QARTOD is not implemented uniformly across the regions, this is something we will deal with in our way I guess. I am not a fan of flagging data bad because one test did not pass our made up thoughts on how data should look. Of course, I am not saying it is not worth looking at, but it concerns me.
  • Climatology tests are also risky, and only should be applied in very very very well understood and oversample regions. Especially during Climate change...in Alaska, we would have missed the BLOB if we had used climatology QC flagging. It is the values outside the climatological mean we want...so I guess flagging them is good, but never fail.
  • Setting thresholds...dos and don'ts. We had a person from NDBC call us when I was working in Maine...told us our buoy data was failing QC and all data being thrown out. It failed climatology tests, based on one mooring data 100 miles away from the 70s...in one of the most dynamic locations in the Gulf of Maine. This was a gross breach of ignorance that I will never forget. I am glad you have experts chiming in, but it can still be risky, so you need tests for the tests, and should never fail Climatologies.
  • Those are my wave buoys 45022 and 45024... It's been an issue to output useful information on wave height and period in the pre qartod era and to various data outlets. I've been sending repeated values (for up to half an hour) of wave height and period when those fail my range tests on the buoy. That's useful information for boaters since especially at low wave heights the wind conditions they don't change quickly... but now that ay trigger a flat line failure. It's good to see the test criteria being used and understand the new system.
  • I'm curious to hear whether you're integrating the flags (at least the roll-up primary flag) into your core user products where you present user-friendly data plots. I think you said you're not doing that yet.
  • I got a little lost in all the acronyms. Maybe take some time to describe items named with acronyms.
  • It'd be good to have a list of all valid xml tags that GLOS can accept (github?).
  • Some of the test parameters might be done by lake if not by station to handle differences in productivity and conductivity.
  • The -998 failure code being used in the HABS portal is a problem because it's killing the autoscaling of plots

GLOS

Clone this wiki locally