Quality Assurance/Control of ADCP Measurements

Provide justification for throwing out a measurement (in field notes if in field and during post processing). The person reviewing your measurement may not have been in the field with you! Provide justification for changing a threshold during post processing.

Here are a few notes from a USGS webiner I attended that I want to pass on to you:

• Rating a measurement is completely subjective, few ratings get an excellent rating (within 2% of true discharge).
• Use the COV (the difference from the mean std/avg in F12 Q summary/WinriverII) to establish a base uncertainty rating (see emily for details). Take the COV/[(# transects)^(1/2)] for your base uncertainty rating.
• Consider the following when REVIEWING a measurement: 1. Variability of transects, 2. patterns in transects, 3. Accuracy of top and bottom extrapolation, 4. Edge extrapolation, 5. Quantity and distribution of missing data, 6. Quality of boat navigation
• Also consider the following when RATING a measurement: 1. data review, 2. hydraulics, 3. site conditions, 4. statistics (COV), 5. Judgement
• Top extrapolation seems to be an important issue (as we suspected), particularly in shallow streams. We should check the different extrapolations (we first assume it is a log profile (power fit)) especially if windy conditions affecting the surface velocities. Recommend try differnent methods of extrapolation and compare final Q. Note how much percentage of flow is in the top discharge estimate. May not be necessary to fully review each transect for extrapolation method, but pick 1 or 2. Assume power fit unless the data tells you otherwise.
• If you have missing data in your profile, it might be okay, look at the distribution/quantity of the missing data in the velocity contour profile to determine the extent of the problem.
• Think about the percent of flow for the error you are looking at. If it is a small percentage of the flow (i.e. edge estimates or a small number of missing ensembles) don't worry too much. If it is large (i.e. top extrapolations, compass problems, overall bad stick ship track) then worry about it more.
• The USGS recommends less than 1 degree total error for the Compass Evaluation. (No mention of error during calibration). If you have greater than 1 degree, you might have trouble in the calibration. (?) We should ask RDI about this, their guidance recommends 2 deg for calibration and evaluation.
• Compass calibration is needed for a loop MB test and for GPS. I *think* this means that you should only do a stationary MB test for a Streampro w/out a compass. Our new streampro will have a compass so we can do the loop test from here on out.
• Do reciprical transect pairs. Directional bias can be seen with use of GPS. If there is a directional bias w/ GPS it may be due to a compass error. You might be able to adjust the magnetic variation during post processing to correct for the compass error.
• Excellent ratings are rare.
• If you have bad (missing) bottom track data during your loop test, you can't use it to correct data in moving bed situation (this was in our example during the training last week when there was some missing data during the MB test, probably due to loss of bottom track).
• Use LC (loop correction) or SMBA (stationary moving bed analysis) to check for moving bed rather than your own calculations (more room for mistakes if you do it yourself). You should have this software on your computers from the last training.
• If you have read this, thanks for reading, I am impressed.

Have the following screens open during the measurement and the data QA/QC as needed:

• Discharge summary F12
• Ship track showing a good straight path for BT, GGA, and VTG track, look for irregularities, GPS/compass calibration issues, moving bed issues. Subsection any zigzags if necessary.
• Stick ship track, look for velocity errors (use the up/down arrow to change the depth). These may be screened out using thresholds.
• Earth velocity magnitude, look for gaps/white stripes in data, bad or no bottom depth, bad bins, etc. Gaps indicate a loss of bottom tracking. Look at edges and see if there are good bins at edge. This is needed for estimating the edge discharge. Subsection if necessary but make sure the distances are still okay.
• Water Speed vs. Boat Speed time series, look for spikes and dropouts, make sure boat speed is less than water speed. At the edges, the boat speed should be at rest.
• Heading/Pitch/Roll time series, look for spikes, etc. Pitch and roll should be close to 0 and less than 5 degrees on average.
• Tabular – Compass calibration
  • review BMG-GMG mag (bottom track – GPS track dist made good) should be close to zero, otherwise moving bed or GPS errors or compass is not calibrated
  • BC/GC should be close to 1
  • GC-BC should be close to 0 or 360 degrees.
• Tabular - GPS
  • Number of satellite changes should be low
  • Delta Altitude should be low, less than 3.5 m is acceptable
  • HDOP should be less than 4, around 2 is good.
  • Delta HDOP should be low
  • If many items appear red, GPS-derived velocity quality may be degraded.
• Earth Error Velocity (go to view – graphs – contour – velocity then right click on the screen and pick “data selection”. Pick Earth Error Velocity). Look for a very random distribution of error. Patterns or white stripes may indicate problems or possible velocity errors.
• Earth Projected Velocity (go to view – graphs – contour – velocity then right click on the screen and pick “data selection”. Pick Eearth projected Velocity). Press F2 to set the projected velocity to look for reverse flows, etc.
• Discharge Profile (to review estimated areas) (press F2 to average 5-10 ensembles and scroll through the profiles to see if default power fit extrapolation technique is okay).
• Press F3 for processing properties. Review the commands, edge estimates, offsets (such as transducer depth), thresholds. Check the thresholds. Thresholds may be adjusted to screen out bad data. In general we leave it as the default but adjustments are needed for streampro IPAQ measurements because the defaults are wrong. Adjusting thresholds is not recommended to fix your bad data…… getting a better measurement site is best when possible.
• Press F9 for system parameters -- the ADCP that was used, the water mode, bin size, etc.
• Press ctrl-F9 to average ensembles (during review)
• Press ctrl-G and ctrl-B to go back and forth between bottom track and GGA reference