search

frank-engel-usgs / VMT

The Velocity Mapping Toolbox
Other
19 stars 12 forks source link

Time computation and VMT Excel Output #3

Closed frank-engel-usgs closed 7 years ago

frank-engel-usgs commented 7 years ago

When applying the Flip XS (auto vs manual), the time stamps in the excel file and UTM coordinates are not being handled correctly.

For example, if I export an excel file with the auto XS feature on, I get the boat starting at one end of the line (see attached plot) and if I manually set the start bank, I get the boat starting at the other end of the line. In reality, the boat started offshore at 17:05 on 9/1/15. If set to auto, the flow direction forces the boat to start on the left bank (or nearhshore in this case) at 17:05 on 9/1/15. Therefore, depending on what I choose, my boat can be at two different points at the exact same time.

The time stamps and coordinates must be maintained and sorted as a group to ensure that the time stamp for the observation is accurate. This will especially affect modelers trying to apply these data to a model based on time stamp. In addition, if applying a tide file after export to Excel, we need to ensure the time stamp and bed elevation measurement are true to the data.

The simple work around is to reverse the time stamp vector when writing to excel, but the problem is more pervasive since the correct timestamps are not being maintained with their respective UTM grid nodes in the V structure as stationing is assigned across the channel (I assume).

I have not dug into this and I am not sure I will do that today, but I wanted you to be aware of this.

We might want to find a way to concisely present the issue and post it on the forum as an FYI.

image

frank-engel-usgs commented 7 years ago

I was able to verify that this is a processing error. I tested with line 005 from the WabashVermillion code test data. When viewed in WinRiverII, the first ensemble occurs on the channel right bank at 09:48:58: image The last ensemble occurred on the channel left bank at 09:52:30: image

In the VMT excel output, the first grid node is on the left bank (this is correct, for the Auto stationing method), but it's time is 09:49:23. Similarly, the last grid node is on the right bank, with a time of 9:52:18.

When I switch the stationing method to manual and set the start bank to right, the Excel output shows the correct UTM starting position for the first grid node, and a timestamp of 9:48:58. The last gird node is on the left bank, with a time of 9:52:14.

So, to fix this bug requires closely examining that VMT is not seperating the ensemble timestamps from the positions when it does the mapping to the mean cross-section (MCS). My guess is that changes will be made to VMT_MapEns2MeanXS and VMT_GridData2MeanXS.

frank-engel-usgs commented 7 years ago

The issue was caused in VMT_GridData2MeanXS. The Time was being computed from the ADCP raw times, but not sorted in the same manner as the rest of the data. To fix, I let VMT sort the times as determined by A(zi).Comp.vecmap, which is computed in VMT_MapEns2MeanXS subfunction InterpBadGPS.