Front Finder!

[sophieclayton]

frontfinder - a package to detect fronts in ocean data

Summary

Develop an easy-to-use package that will ingest ocean data and find fronts (e.g. large gradients in density, temperature, chlorophyl...).

Personnel

Sophie Clayton/TBD

Data sets and infrastructure support

This could be tested out using glider/AUV data as a first step? Would also be useful to use

The problem

Although there are some handy off-the-shelf remote sensing products that can be used to locate fronts in SST and Chl (NOAA Chl front magnitude product), it is a recurring problem for oceanographers to identify fronts from in situ data (e.g. underway TSG, glider sections, SailDrone, shipboard ADCP...). The goal of this project would be to provide a solution for better identification of fronts from in situ ocean data (possibly also making use of remote sensing products). There has also been a focus on finding fronts in sea surface data, but with depth-resolved sections we can get a better view of the structure of fronts.

Maybe combine surface and depth-resolved data?

Existing methods

There are many existing algorithms for front detection, particularly from satellite data.

• simplest is using a gradient criterion, e.g. as implemented in this paper: Levy at al., 2015
• edge detection implemented in a nice notebook put together by @ocefpaf

[lidefi87]

New repository started by OHW Australia participants: https://github.com/oceanhackweek/OHW22_proj_front_finder.git

[vsilvafelipe]

Hi, @sophieclayton. I am interested in the project. I discussed with @mayajakes and we thought of using the peaks in |SSH gradients| to set a region where fronts are found. It works well for regions where there is a surface-to-bottom current (e.g., the ACC) and the altimeter SSH gradient captures the surface-to-bottom density gradients. Then, we thought of using the in situ observations (Saildrones?, ARGO?) to evaluate the hydrographic information within the area of high |SSH gradients|. In this case, we would not use the profiles out of the regions of strong |SSH gradients|. If the hydrography structure is consistent in different coordinates of peaks in |SSH gradients|, it could return subsurface characteristics of fronts. What do you think?

@NickMortimer suggested us to start a repository with these ideas. @lidefi87 helped us with that.

[sophieclayton]

that sounds great!

[vsilvafelipe]

About using the program for any variable, I think we could use normalized gradients. In this case, the user would set the limit for the position of the front in percentage instead of a value for |SSH gradient| or |SST gradient|. What about? @sophieclayton @mayajakes @MackenzieBlanusa

[mayajakes]

We have created a slack channel called #frontfinder-project within the Oceanhackweek slack, so if you are interested in this project feel free to join!