search

Coastal-Imaging-Research-Network / cBathy-Toolbox

Routines needed to run cBathy + demos
https://github.com/Coastal-Imaging-Research-Network/cBathy-toolbox/wiki/cBathy-User-Manual
GNU General Public License v3.0
25 stars 23 forks source link

Anomalous Shallow Regions #18

Closed SBFRF closed 2 years ago

SBFRF commented 7 years ago

Hello all, I've found an issue in the cBathy output depths at the Duck NC location. This first appeared to me in a data set that was recently reprocessed to remove some of the effect caused by the overlapping camera seam effect that was evident at the duck location. In the attached image I've plotted the old version (with the seam) and the new version (with out the seam) that has this anomalous depth patch. The date in which the sandbar first presents itself is on 2015-10-15 1959 UTC. It sticks around for approximately a day before the kalman filter is willing to let go of the object.

newlyreprocessedcbathy

This shows a 0 m NAVD88 depth in 4 surrounding 4 meters of water at approximately 400 meters in the cross-shore and 1000 meters in the alongshore.

I've also found another example of this at Duck, with the date 2016-07-19 1559 UTC, unfortunately I don't have the capability to plot this right now. There are two features at this time, on the south property, both of which stick around for approximately a day. These were at (300m, -300m) and (350m, 200m).

bergsmaE commented 7 years ago

Do you (or somebody else) know which technique has been used to reduce the camera seam issue? I think this is a very unlikely result with the solution provided in Bergsma et al. (2016).

I know that I should be able to know how to get this by now but can you send me somehow the stack and associated geometry?

KateBrodie commented 7 years ago

I believe @RobHolman and @jstanleyx reprocessed this data, though I'm not sure what technique was used. I think it also involved increasing the smoothing length-scales, which may be contributing to this, since it is a day with very short period waves. @RobHolman has more insight on this, I think.

I also notice in these re-processed bathymetries that the sandbar appears much deeper than in the original data set (perhaps also owing to smoothing?).

@SBFRF, I will also note that the "seam" effect is still pretty clearly present in the offshore portion of the domain even in the newly processed bathymetry.

mpalmsten commented 7 years ago

@bergsmaE, would you be willing to push your fix to the cBathy seam problem to the development branch and do a pull request? It would be cool to re-run these data using your code.

@RobHolman did you increase the smoothing length-scales on the reprocessed data? How much?

@KateBrodie, good observation about the deeper sandbars.

bergsmaE commented 7 years ago

@mpalmsten Yes, no problem. I need some time though to test the routines.

RobHolman commented 7 years ago

@mpalmsten Lx Ly were updated to 3 dx dy for recent runs that Spicer was commenting on. That said, I would need to talk to @jstanleyx about how many runs were reprocessed in this way. This brings up the question of how well we are documenting changes we do when we re-process. We do capture the params structure so I know about the smoothing scales. But I'm not sure if we can tell whether the seam problem is fixed.

@bergsmaE We should compare how we handled the seam problem. It was a pretty easy fix but we should compare. @KateBrodie @mpalmsten I just noticed that the current CIRN version does NOT have any seam problem fix.

mpalmsten commented 2 years ago

Thanks @SBFRF for posting this issue. The problem is addressed in cBathy v2.0, so I am closing this issue.