```html Using Ocean Eddies to Route a 1 MPH Seastead (Caribbean focus)

Routing a ~1 MPH Seastead by “Riding” Caribbean Eddies

Scope & assumptions: You described ~1 MPH continuous propulsion relative to local water (≈0.87 knots) and asked about exploiting mesoscale eddies/current features near Anguilla and for a Caribbean loop. This is an operational-oceanography + routing problem. The notes below are practical/engineering-oriented and not a substitute for professional marine routing or safety planning.

1) What sources predict eddies into the future? How far is reliable?

Primary forecast sources (models)

“Eddies” in products like RTOFS/HYCOM are typically mesoscale features (~10–200 km) resolved by high-resolution ocean circulation models. The usual sources:

NOAA RTOFS (often HYCOM-based). Provides short-range ocean forecasts (currents, temp, salinity) for operational use.
Good entry point: your link and NOAA/NCEP RTOFS pages (currents are derived from model velocity fields).
HYCOM global/regional outputs (HYCOM consortium; various runs and archives).
Copernicus Marine (CMEMS / Mercator Ocean) global and regional ocean analysis/forecast products with currents and sea level anomaly.
ROMS-based regional models (university/regional centers) can be very good near coasts, but coverage varies.

Observation-based “nowcasts” (to detect eddies today)

Satellite altimetry (sea surface height / anomaly) is a key way eddies are identified and tracked. AVISO/CMEMS products are commonly used.
SST (sea surface temperature) imagery can reveal fronts/eddies (but clouds can limit it).
Drifters/Argo floats (less direct for real-time routing, but useful for validation and climatology).

How far ahead are eddy predictions “reliable”?

In practice, eddy existence and general region can be somewhat predictable, but exact position, phase, and the best “lane” degrades quickly.

Forecast horizon	What’s typically usable for routing?	Why it breaks down
0–3 days	Often quite usable for tactical routing (which side of an eddy, where a jet is strongest).	Still sensitive to assimilation timing, cloud gaps in SST, local wind bursts, etc.
4–10 days	Sometimes usable for strategic planning: “a favorable feature likely persists in this corridor”.	Mesoscale phase errors grow; eddy centers can drift tens of km or more.
10–20+ days	Usually only broad-brush (climatology-like): prevailing currents, seasonal tendencies.	Chaotic eddy dynamics; small errors compound into large track/strength errors.

Rule of thumb: treat mesoscale eddy routing as a “continually re-optimized” problem with frequent updates (daily or every few days). Don’t expect a single plan made weeks ahead to stay optimal.

2) Software/packages/algorithms that work with eddy forecasts (and open source options)

Core idea

You have a vehicle with limited speed through water, moving in a time-varying velocity field. The standard approach is:

Ingest a gridded current forecast (u,v) over time (NetCDF typical).
Simulate motion (advection + your propulsion vector, with constraints).
Optimize route for minimum time / maximum progress / energy, re-run as forecasts update.

Open-source building blocks

OceanParcels (Parcels) – Lagrangian particle tracking in Python; widely used in oceanography.
Use it to simulate many candidate trajectories through forecast currents.
OpenDrift – Lagrangian drift modeling framework (Python). Often used for oil spill/drift, but applicable to “vehicle in currents” with custom forcing.
NOAA GNOME (General NOAA Operational Modeling Environment) – drift modeling; more spill-focused but includes current/wind drift machinery.
xarray + netCDF4 + dask – standard Python stack for efficiently handling large gridded ocean model data.
py-eddy-tracker / eddy tracking toolkits – there are open-source libraries that detect/track eddies from sea level anomaly fields (commonly using Okubo–Weiss or closed-contour methods). These help identify eddy centers/boundaries rather than directly routing.

Routing/optimization algorithms that are commonly effective

Time-dependent shortest path on a grid: discretize space; cost to move depends on currents at time t; solve with variants of Dijkstra/A*.
Model predictive control (receding horizon): optimize 1–5 days ahead, execute part of plan, then re-optimize with new forecasts.
Level-set / front propagation methods: compute “earliest arrival time” surfaces in a flow field (good for minimum-time problems).
Monte Carlo / ensemble routing: run many perturbed forecasts/initializations and choose routes robust to uncertainty.

Practical recommendation: For a DIY project, a strong path is: xarray (read currents) + Parcels or custom integrator (simulate) + a search algorithm (A* / dynamic programming) + frequent re-plans.

3) How well can “eddy surfing” work for a 1 MPH platform in practice?

It can help, but the ceiling is limited by (a) how strong/coherent the favorable flows are where you are, (b) how often you can stay in them, and (c) how much detour/time you spend “hunting” for them.

Typical magnitudes: Many mesoscale features have surface speeds on the order of ~0.3–1.0 knots (0.35–1.15 MPH), with jets/edges sometimes higher. Some narrow currents/straits can exceed that.
With 1 MPH through-water speed: you can materially change your track relative to the water and “choose” lanes—especially if you plan early and re-route often.
Realistic average boost: If you can frequently find +0.5 to +1.0 MPH favorable set in your desired direction, your speed over ground might average ~1.5–2.0 MPH for long legs. But it will not be continuous; there will be periods of cross-currents, weak flow, or adverse flow.

Operational reality: Expect “stop-and-go” optimization: sometimes you make great progress for several days, sometimes you slow down or wait/reposition to catch the next favorable feature.

4) “Eddies can’t go through land, so we don’t have to worry about being pushed into land,” right?

No—being set toward land by currents is still a real hazard. Even if an eddy “doesn’t cross land,” its flow field can have onshore components, especially where currents interact with island shelves, headlands, banks, and boundary layers. Also, near islands the flow can accelerate in passes and around capes, creating strong localized sets.

Key points:

If the current toward shore is faster than your maximum speed away from shore, you can still be forced closer (net drift toward land). A 2 MPH onshore set against a 1 MPH capability is a losing battle unless you can exit sideways into weaker flow before you get too close.
“Motor away” depends on geometry and margin. With plenty of sea room you can often crab out of the strongest flow; close to shore, you may not have time/space.
Wind and waves matter a lot. You said “ignore wind,” but for a platform with large above-water area, wind-driven leeway can dominate at times and can push you shoreward regardless of current direction.

Practical mitigation: maintain large “coastal buffers” (tens of miles, sometimes more), use conservative no-go zones near islands/shallows, and treat near-coast routing as a separate, higher-risk mode.

5) Estimated time for a loop around the Caribbean Sea using eddies

What “loop” means (definition)

There are many ways to define “around the Caribbean.” A reasonable “family seastead loop” that stays mostly in open water but traces the basin could be approximated as:

Start: Anguilla
South/west along the Lesser Antilles arc (keeping offshore)
Across toward the ABC region / or north of Venezuela (depending on avoidance)
West toward Central America/Yucatán region
North/east via channels around Cuba/Bahamas region
Back to Anguilla

Distance estimate

A coarse but serviceable estimate for such a “basin loop” is on the order of:

~3,000 to 4,000 nautical miles (≈3,450 to 4,600 statute miles)

Speed over ground scenarios

Your propulsion: 1 MPH through water (≈0.87 knots).

Scenario	Average favorable current component (net)	Average speed over ground	Time for 3,450–4,600 miles
No meaningful help	~0 MPH	~1.0 MPH	~144–192 days (4.7–6.3 months)
Moderate eddy help	~+0.3 to +0.6 MPH	~1.3–1.6 MPH	~90–148 days (3.0–4.9 months)
Strong, consistently exploited help	~+0.8 to +1.0 MPH	~1.8–2.0 MPH	~72–106 days (2.4–3.5 months)

My practical estimate: if you are actively routing and re-optimizing with forecasts, ~3 to 5 months is a reasonable planning range for a “Caribbean loop,” with occasional “fast weeks” and “slow weeks.” Getting reliably down near ~2.5–3 months would require unusually good alignment of currents with your chosen direction, plus tolerance for detours and waiting.

Clockwise vs counterclockwise: Either can work, but the “best” depends strongly on season, where eddies are shedding, and how you define the loop. In practice you would choose direction opportunistically based on the next 1–2 weeks of forecast flow.

6) Are there usable eddies in other places (South Pacific, Mediterranean, E. South America)?

Yes—mesoscale eddies are common in most oceans. What changes is (a) their size/strength, (b) how easy they are to observe/forecast, and (c) how much land/shallows constrain you.

Mediterranean

Many eddies and fronts, often smaller spatial scales than open ocean.
Forecasting can be good (strong modeling community; CMEMS has dedicated Mediterranean products).
Constraints are tighter: more coasts, shipping lanes, winds (e.g., Mistral, Meltemi) can dominate surface drift.

South Pacific

Plenty of eddies, but distances are huge; “eddy surfing” helps but doesn’t remove the basic scale problem.
Trade-wind effects and wave climate become major operational constraints for a slow platform.

Eastern coast of South America

Brazil Current (southwest Atlantic) is a major western boundary current with eddies/meanders; can be strong and useful (or hazardous).
North Brazil Current / retroflection eddies (farther north) are energetic and can transport water rapidly offshore—again potentially helpful but requires cautious routing.
Big caution: boundary currents can run fast near the shelf break; being “pinned” toward a coast/shelf by wind + current is a credible risk for a slow craft.

7) Practical next steps (if you want to build this into a website/tool)

Data pipeline: choose a primary model (RTOFS or CMEMS), download gridded u/v currents (NetCDF), subset around your AOI, update daily.
Trajectory engine: integrate motion with constraints (max through-water speed 1 MPH, optional “no-go” polygons near land, optional wave/wind leeway later).
Optimizer: compute minimum-time paths to waypoints using time-dependent A* (or level-set), re-run daily.
Uncertainty: run an ensemble of slightly perturbed currents and pick routes that are robust, not just fastest in one forecast.

If you tell me: (1) your desired “loop” definition/waypoints, (2) how far offshore you want to stay, and (3) whether you can pause/wait, I can provide a more concrete distance estimate and a routing approach you can implement (including a simple architecture and pseudocode).

Links (starting points)

NOAA RTOFS (operational ocean forecast currents): search “NOAA RTOFS NetCDF currents”
HYCOM data server/archives: search “HYCOM data access”
Copernicus Marine (CMEMS) forecasts and altimetry: https://marine.copernicus.eu/
OceanParcels: https://oceanparcels.org/
OpenDrift: https://opendrift.github.io/
NOAA GNOME: https://response.restoration.noaa.gov/gnome

```