Thoughts on the future of the sound speed uncertainty project

I'm reviewing my notes from my meeting with Kris Holderied and having a few insights. Kris was suggesting that we take transects with the AUV or MVP from which to create spatio-temporal gradients that describe the oceanographic properties of the water mass.

My first thought was, "Aren't we doing this already?" a la the ROMS/AUV paper. So I went back and reviewed the paper and realized that the problem has only been looked at with the assumption that we are collecting discrete casts and interpolating between them. With AUVs and MVPs we have the ability to collect continuous data which gives us much, much finer resolution for interpolation.

Much of the ROMS/AUV paper focueses on "detecting" when a discrete cast has occured, but that misses the point. Why try and detect a discrete cast and then discard the rest of the data, when we could use that data as part of the interpolation?

And for that matter, why do we continue to down sample all of the sound speed profiles to whole meters. There's a lot more data in there that could be used (and it's highly likely that we have the processing capabilities to handle the full data now, whereas we didn't in 1987 or so when Velocwin was first written for DOS).

My point is that there are still a number of NOAA-specific operational assumptions which are driving this research, and we need to filter out as many of those as possible, before moving forward.