Marine Data Literacy 2.0

Providing instruction for managing, converting, analyzing and displaying oceanographic station data, marine meteorological data, GIS-compatible marine and coastal data or model simulations, and mapped remote sensing imagery

 

 

 

 

Home > 4. Ocean Data View > 4.19 ODV Flags

4.19 Editing Quality Flags for Sample, Station and Cruise Data in ODV

  • Exercise Title:  Editing Quality Flags for Sample, Station and Cruise Data in ODV

  • Abstract:  In this exercise you'll learn more about how to edit ODV data collection to "flag" suspicious or known bad data with markers that allow you to avoid them during display and analysis work.  Quality flag edits can be made at the sample level, the station level or for entire cruises.  The data are not lost, but are still included in the collection with changed flags, so the edits are not permanent if events later indicate they should be reversed.

  • Preliminary Reading (in OceanTeacher, unless otherwise indicated):

  • Required Software:

  • Other Resources: 

  • Author:  Murray Brown

  • Version:  7-13-2014

1.  Open the Liberia bottle data collection.  This shows the familiar TS diagram, but you can begin with any view.
2.  You can do any of the quality control steps below from just about any of the views of you data collection, due to the infinite flexibility of ODV to connect displays with data properties.  Below we'll use a different view, but you could just as easily work from this view...it's completely up to you.  The important thing is to display the data in meaningful ways, and look for discrepancies or oddities it identify possible quality issues.
3.  The author recommends a particular way of showing the data in a map, very similar to SURFACE MODE, but created slightly differently.

Select VIEW > DERIVED VARIABLES.  Open the METADATA tab and select LATITUDE.  Move it to the DEFINED side with the ADD control.

4.  Also ADD the LONGITUDE variable.  Then click OK at the bottom right.
5.  Back on your data plot, right-click and select X-VARIABLE.
6.  Select LONGITUDE and click OK.
7.  In the same way, select LATITUDE for Y.
8.  Your view should look something like this.
9.  On the small station map, right-click and select PROPERTIES.
10.  Open the DATA/TIME tab, and select SEASON > JAN 01 TO MAR 31.

Then click OK at the bottom right.

11.  Right click on the data view itself, and select SAMPLE SELECTION CRITERIA.
12.  Select the RANGE tab, and set the acceptable range for DEPTH to be 0 to 5.

Also check the box APPLY THESE SAMPLE SELECTION CRITERIA GLOBALLY just to make sure.

13.  The author recommends this way to isolate the "surface" data, rather than the use of a SURFACE MODE map, such as TEMP AT DEPTH=0, because it provides absolute control over what you're working with.
14.  Similary open the QUALITY tab, and select 0: ACCEPTED VALUE.

Then click OK to apply these two criteria.

15.  There are changes, but not very marked.
16.  Now as you did above for the X and Y variables, select TEMPERATURE for Z.
17.  This is the JFM 0-5 m temperature map, for Liberia stations in the WOD, using only accepted-quality data.
18.  Before we do anything else, lets look at the range of data values.  Right-click on the view and select EXTRAS > STATISTICS
19.  Take some time to study these possibilities.  Then click on Z HISTOGRAM to see the temperature value range.
20.  And this is the range of temperature values.  The very lowest values (on the left) look suspiciously low may be strongly effecting the analyses, so we'll try to eliminate them.

The lowest value is 12 degrees cooler than the median, a very unlikely value!

21.  Using ZOOM with the main view, you can expand the map to the region you see here.  There are 2 very low values, as indicated by the blue and purple colors.  These might be responsible for the "tail" on the left of the histogram.
22.  If you click on the left point, these metadata are displayed.  The cruise is WOD13_FR008518.
23.  If you click on the right point, the same cruise is identified.  Evidently these folks had some trouble with their thermometer data during this cruise.
24.  Go back to the left point.  Right-click on the actual temperature value itself, 16.97.  Then select EDIT DATA.

NOTE:  ODV is almost limitless in the number of ways you can edit data, and all three of these can work.  The author just picks the easy one, but it's up to you.

25.  This EDIT DATA window opens.  Select CHANGE QUALITY FLAG.
26.  This opens the native quality flag list from the WOD data.  You can study these on your own time later.  Right now we just want the easiest way to flag this measurement as suspect, due to its great difference from the surrounding data.

Select RANGE OUTLIER as the easiest way to do that.  Then click OK.

27.  The colored point has become a tiny dot, marking the location.
28.  Now select the right point, and go into the EDIT DATA window.  It would seem that the two values shown are too cold to be reasonable for this location, so you can select them both.

Then select CHANGE QUALITY FLAGS as you did above the other station.

29.  When you are done with this second station, it also becomes only a tiny dot on the map.

This means that both stations are still in the database, but they are not included in displays or calculations or exported subsets.

30.  So that's how we can work with single samples (the left point), stations (the right point).  Now we can move on to whole cruises.  This would seem to be a drastic move, but as you'll see below it can make a huge difference in the map views.
31.  To get back to the full data map, right-click on the small station map, and select FULL DOMAIN.
32.  Here we are back to the full domain.  Many students have noticed the 3 north-south lines of stations that seem to appear a little warmer than the surrounding stations.
33.  The effect of these stations is obvious, especially when gridding the data, as in the exercise 5.2 Gridding Point Shapes in Saga.  The figure is here is from that exercise.
34.  If you select stations along the far western line, you'll find that most of them have these metadata.  The cruise is WOD13_US00090.
35.  And if you search around in the middle and eastern lines, you'll find that they both came from WOD13_US000998.
36.  OPTION:  You could go to the metadata catalog for these Liberia data that you captured in the exercise 3.2 Managing/Repatriating Area-Specific Marine Metadata from the World Ocean Database (WOD) to see if there is anything strange about these cruises, or if the locations might be wrong.  This is just a remote possibility, but it never hurts to know more about data sources when questions are raised about validity.  The author reports that they appear to be completely in order, and part of a larger survey off western Africa.
37.  To flag these data as questionable, right-click on the station map and select STATION SELECTION CRITERIA.

Beside CRUISE LABEL, write in the full cruise ID's for the questionable data.  There is a special syntax for multiple entries, in ODV's instructions.  individual cruise ID's should be separated by || as you see here.

Then click OK.

38.  Now you see only the stations from these cruises.
39.  The presence of the cooler stations to the north makes this a very complicated situation.  We will continue here to flag all the station as questionable, but you should study the real situation in depth to see if you agree with this for all of them.  The author just wants to demonstrate the general method here, but he warns you that the northern stations might be perfectly OK, while the southern ones probably are not.  But if there is an indication of data problems for any variable at most stations of a cruise, then it is a good bet that problems persist with that variable for the other stations.
40.  Now that the "problem" cruises have been isolated, we can work on the flags in batch mode.

Right-click on any temperature value for any visible station.  Then select ASSIGN QUALITY FLAG > ALL SAMPLES OF VALID STATIONS.

NOTE:  "VALID STATION" is ODV's way to identify all currently visible stations.

41.  ODV warns you that you're about to make a major change.  If you want to continue, click YES.
42.  Here's the temperature samples map after the change.  All the stations in the subject cruises have been changed to tiny dots.

You can easily see that the whole area seems to have more evenly distributed temperature values, so it seems our edit has had a positive impact.

43.  And here we've right-clicked on the map and applied PROPERTIES > GENERAL > GRIDDED FIELD > QUICK GRIDDING.  You can see that the north-south lines have been removed.
44.  ODV has kept track of all your changes, and the current collection has already been updated.  You can exit the collection at any time (perhaps saving the view) and come back to this state (values and flags) at any time.