Argo floats in shallow waters

[j-atkins]

I have noticed erroneous output from the Argo floats when deployed in shallow waters. See for example the screenshot below where I've deployed an Argo float in shallow (< 200m) waters, but it continues sampling down to 2000m depth, sampling 0s values of temperature.

I haven't done a full diagnosis yet but I suspect it is something due to the fact that all expedition.yamls will default to the following Argo config:

argo_float_config:
    cycle_days: 10.0
    drift_days: 9.0
    drift_depth_meter: -1000.0 # <-- see here deep default
    max_depth_meter: -2000.0 # <-- see here deep default
    min_depth_meter: 0.0
    stationkeeping_time_minutes: 20.0
    vertical_speed_meter_per_second: -0.1

Where the drift_depth_meter and max_depth_meter fields default to very deep values.

If Argos shouldn't really be deployed in shallow waters, then I think we should introduce blocks on deploying them when e.g. bathymetry < 2000m.

But, if it's good to have the flexibility to deploy shallow, then we need to think a bit further. For example, should the max depth and drift depth be determined from the bathymetry for the location, overriding the config if the default values are too deep? With a warning?

Keen to hear your thoughts @erikvansebille and @ammedd!

I notice there has been previous discussion about shallow Argos (#26), which could be related.

[erikvansebille]

I'm not sure actually how Argo floats avoid colliding into the seafloor when the bathymetry is shallower than 2,000m. From the "High speed communications" section at https://argo.ucsd.edu/how-do-floats-work/technological-innovations/, it seems they don't have an avoidance mechanism? Maybe they just rest on the seafloor?

Perhaps reach out to the Argo team to ask what is most realistic?

[j-atkins]

Hmm I see - interesting! I'll look into this further and indeed get in touch with the Argo team if I don't get anywhere soon.

[j-atkins]

Update: I have seen from an Issue in the VirtualFleet project that real life Argo floats will usually raise themselves by 50dbar if they ground on the sea floor (which is a feature that VirtualFleet then added).

In any case, I have emailed the Argo team to ask for their insights. I imagine they'll confirm something similar. I'll add a feature along these lines to the VirtualShip Argo floats.

[j-atkins]

Grounding behaviour now added in #263.

I will also be speaking to the Argo team next week to discuss further realistic behaviours, so I'm keeping this issue open for now in case there are further changes to discuss.

[ammedd]

Cool! Please invite me for the discussion with the Argo team. I'll also update you on my talks with NIOZ about Argo floats. #26 was indeed related. I'll close that and refer to this issue.

[j-atkins]

I met with the Argo team earlier this week. I will list the minutes below:

Grounding behaviour

• The grounding behaviour of Argo floats is model dependent but tends not to be overly sophisticated (they will often just bob along the sea floor if they hit it)
• The +50m depth behaviour as implemented in Argo floats in shallow waters #263 is a reasonable approximation of real life.
• As a final resort, the real-life floats are programmed to simply return to the surface and repeat the cycles if encountering problems. This is something we could think about introducing should we come across unexpected issues in the future.

BGC Argo floats

• Which sensors are used on BGC Argo floats is still an open and actively researched area.
• Oxygen sensors are the most mature and commonly used.
• Other 'operational' sensors include chlorophyll, optimal backscatter, ph, nitrate, irradiance (a float will very rarely have all 6 though)
• Phytoplankton, acoustics, POC sensors are currently experimental and not IOC approved.
• The Argo community are keen to do experiments with BGC Argos, e.g. testing distribution strategies, where to deploy etc. This is very expensive though so there's interest from their side in VirtualShip use once we've incorporated BGC!
  • Definitely worth alerting them once we've done this.
• On a similar note, it is common to do CTD casts with Argos in the experimentation, so VirtualShip's multi-instrument approach is helpful here.

Outreach

• Argo team share resources for education, research etc. They'd be happy to point to VirtualShip and promote.
• See point above about interest in using VirtualShip for BGC Argo experimentation.

Deep Argos

• Deep Argo is an ongoing mission but has less funding.
• Argos reaching 4k depth resemble the traditional shape, those reaching 6k though are quite different (glass).
• An application is bathymetry mapping, simply by seeing when they hit the seafloor.

Lifetimes

• Real life Argos often aim for multi year lifetimes (~4-6 years).
• By default, in VirtualShip, we encourage shorter deployments (~months), in part I imagine because of compute time.
• Signposting the ability to extend the Argo lifetime (as is being incorporated in Remove spacetime-region #264) might be a good idea, for those who want to focus on Argo usage in VirtualShip.

[erikvansebille]

Thanks for this update, @j-atkins. Seems like a very useful meeting!