When collecting sonar logs in a grid pattern there will be many points that are measured twice. It should be possible to compute a tide correction function using the time and depth deltas for each of these grid crossings. Assuming that the data is collected over a time period that isn't any more than about 5-6 hours, a 3rd order polynomial is sufficient to give ~5 cm precision. Has anyone tried this?

I'm assuming that the calculations would need to be done outside of ReefMaster. Once computed, it would be possible to generate a table of time vs. tide height. Would it be possible to then import this table into ReefMaster and then use it to correct all points on the grid?

An old Captain I sailed with when I was a young officer once told me "Navigation is the science of approximation" (he later used the same phrase to the Navigating officer, followed by "but you are far too f........ approximate).
While it looks good in the armchair it is probably unrealistic to expect such accuracy in the real world. You are counting on the repeatable accuracy of the GPS to be spot on with each pass. How do you know that your antenna is exactly where it says it is ? In order to do this to centimeter accuracy you need to know that you are at exactly the same point or the depth will be different due to bottom structure inconsistency, not just tide. Then there are waves, squat, etc.
It's a great method on a lake for getting a good average though when reducing tracks to a datum, but its only ever very approximate.

There are a lot of folks who see all the wonderful accuracy figures quoted and go out and buy kit on the basis of it. It's mostly smoke and mirrors, be content with reasonable accuracy at a reasonable price.

T

Although the idea sounds good in principal (and is something we have thought about), I think that in practice it would not prove that useful.
Even on small maps with a high data density it is uncommon to get data points in exactly the same position, so to get any sort of reasonable number of cross-matched data points you would need to define some sort of distance over which you assume that the depth doesn't vary (say, a 5m radius) - but of course the depth will vary across some of those areas, perhaps by more than the tidal influence. Then, as Trevor mentioned, you also need to consider the accuracy of your GPS position, and indeed the accuracy of your depth readings. Even given all this, I still think it is unlikely that you would get a sufficient number of usable data points in a single short track and the number of potential inaccuracies along the way could lead to wildly inaccurate results by the end. Perhaps if you combined this with outlier elimination and some sort of curve fitting to the known tide shape.. but I think it would be a lot of effort to get enough data points to make it work. In any case, if you know the tide for the area it is probably far easier to simply enter a few known times/depths and let RM interpolate the rest.

Having said all that, if you do want to try this then it is easy to import your own tides into RM. Just create a CSV file containing UTC time and the depth offset and import into a new tidestation. Values between supplied points will be calculated using a straight linear interpolation, anything outside of the beginning or end of the supplied times will take the offset of the first or last value.