Two linked tools over the real Pollica meadow. The planner sizes a single survey, drawn straight onto the seabed you are mapping. The fleet constructor turns an area and a budget into a deployable proposal: which sonar, how many boats, how many people, and whether to rent it, own it, or run it as a yearly program.
Vessel runs the track lines. Sonar logs plus a GNSS position track are saved on board.
Logs are corrected for speed and slant range, then stitched into a geo-referenced image.
Overlapping passes are blended into one coverage map of the survey band.
Meadow, sand and reef are outlined. A few ground-truth points confirm what the texture means.
Raw logs, mosaic and outlines are held by the municipality. This is the sovereign record.
Repeat the same lines next year. The difference is the change in meadow extent.
What it actually costs to keep an annual record going for three years.
More boats finish faster, with diminishing returns. Your current fleet is marked.
RTK plus a heading source is the upgrade that matters most. Standard GPS scatters a few meters, which blurs real year-on-year change at the meadow edge. If the point is a defensible record, budget the RTK line.
Every model needs a few confirmed points. A diver or drop-camera confirms that a texture really is seagrass. Side-scan maps the extent, the spot checks give it meaning.
Owning the kit only works with a protocol. Same lines, same settings, same season, every year. That fixed method is what Island Lab supplies and what keeps a community-run survey comparable over time.
Design and philosophy
This is a way for a coastal community to map the seafloor it depends on, keep the record in its own hands, and repeat that record every year so change becomes visible. The first application is Posidonia oceanica seagrass in Pollica, but the idea is general.
Small vessels carry an affordable side-scan sonar along the coast. The data is processed into a georeferenced picture of the bottom. The community holds the raw data, the map, and the outlines. Next year the same lines are run again, and the difference between the two is the story of the meadow.
Around that method sits a planning toolkit. It sizes a single survey, then assembles a fleet against a real area and budget, and turns that into a proposal a mayor can act on: which tool, how many boats, how many people, and whether it makes more sense to rent the capability, own it, or run it as a yearly program.
The document below is the thinking behind the toolkit, written so it can seed a new project folder and so it reads honestly to the communities we would hand it to.
Coastal communities make decisions about the water in front of them with almost no current picture of what is actually there. Seafloor habitat maps, where they exist at all, tend to be one-off academic surveys that are years old, held by an institution far away, and impossible for the community to reproduce. When a meadow starts to retreat, or an anchor field starts to scar it, or a restoration planting needs to be measured, there is no baseline the community owns and no cheap way to make one.
The usual answer is to hire a survey company at a scale and cost that only a national agency can afford, or to wait for a research group to publish. Neither leaves the community with a capability. Both leave the data somewhere else.
The gap we are answering is not a technology gap. Capable side-scan sonar is now cheap enough to mount on a kayak. The gap is that nobody has packaged the tool, the method, and the ownership model into something a comune can actually run and keep.
Data sovereignty is the point, not a feature. The single most important thing this delivers is that the community holds its own record. The raw logs, the mosaic, and the outlines live with the municipality, not with a vendor and not with a distant institution. Everything else in the design serves that. When we show a partner exactly what was measured, what was synthesized, and what is still planned, that moment of clarity is consistently what changes the conversation. The tool is built to make that moment easy to reach.
Repeatability beats precision. For a yearly record, running the same lines the same way matters more than perfect accuracy on any single pass. A community survey that is a little coarse but genuinely repeatable will reveal real change over three years. A one-time survey that is beautiful but never repeated reveals nothing about trend. We optimize for the comparison across years, not for the single best image.
Restoration, not stewardship. The framing for community-facing work is restoration. It is the language the communities we work with use for their own relationship to the coast, and it centers action rather than custody. We lead with restoration, with the meadow, with working alongside the comune, and we keep tech-vendor vocabulary out of the first conversation.
Separate the possible from the probable. A good tool is honest about what a community can actually do versus what is outside its control. Running the survey, protecting a meadow with a no-anchor zone, replanting a patch: these are deployable now. The warming trajectory of the sea is a systemic lever the community cannot pull. Any planning surface we build keeps those two categories visibly distinct so that no one mistakes a wish for a plan.
Every number carries its provenance. Nothing in the toolkit is presented as fact unless it is computed from real inputs. Each visible figure is tagged as computed, as an assumption the user set, or as an illustrative placeholder. Hardware prices are anchored to public list prices. Labor rates, upkeep percentages, and margins are placeholders until a real cost model replaces them. This discipline is not decoration. A single conflated rate can inflate an entire model's framing, and a partner or a grant reviewer is right to distrust any number that cannot say where it came from.
Lead with discovery, not product names. For municipal and community partners the first conversation is about the coast and the problem, not about a platform. Product names come later, after the value is understood. The toolkit is a scaffold for that conversation, not a catalog.
The community runs it, we supply the method. Island Lab provides the tool, the fixed protocol, the training, and the quality control. The community provides the hands and keeps the data. Owning a sonar without a protocol produces inconsistent data that cannot be compared year to year. The protocol is the real product, and it is what keeps a community-run survey honest over time.
The pipeline is deliberately short and legible, so a community can understand every step and, over time, run most of it.
A vessel runs the planned track lines, and the sonar logs are saved on board alongside a position track. The logs are then corrected for vessel speed and for the slant geometry of the sonar and stitched into a geo-referenced image. Overlapping passes are blended into a single coverage map of the survey band. Habitat is outlined on that map, meadow against sand against reef, and a handful of ground-truth points, from a diver or a drop-camera, confirm what each texture actually is. The raw logs, the finished mosaic, and the outlines are then archived by the municipality as the sovereign record. The following year the same lines are run again, and the change in extent is read directly from the difference.
Side-scan reads texture rather than depth, which is why this works for seagrass. Dense Posidonia scatters sound differently from bare sand, so the meadow shows up as a distinct pattern, and that is enough to trace its extent and its boundaries. Ground-truth is what turns a texture into a confirmed habitat, so it is never optional.
There are two realistic hardware paths and one autonomous option, and the toolkit lets a community see the tradeoff rather than hiding it.
The pro-grade path is the Cerulean Omniscan 450 SS, the side-scan built around the Blue Robotics stack. It paints a georeferenced mosaic in real time when the vessel feeds it position and heading, which is exactly the sovereign-record workflow, and it lists from roughly $2,590 per side. Dual-side coverage runs about $5,180 in sonar before the topside computer, positioning, mount, and battery. This is the option to reach for when the point is a defensible, repeatable, self-mosaicking record.
The consumer path is a side-imaging fishfinder such as a Humminbird Helix with MEGA imaging, around $1,200 for the unit. The imagery is excellent and the cost per boat is a fraction of the pro-grade path, but the logs are proprietary and need post-processing in software like ReefMaster to become a georeferenced map. This is the option for the cheapest possible community-scale coverage when a little more processing effort is acceptable.
The autonomous option is the same Omniscan sonar on an ASV. It repeats the exact same lines every year with almost no crew in the water, at a higher rig cost and a higher setup skill. It is the right answer where exact line repeatability and low field labor matter more than capital.
Across all three, the upgrade that most determines whether the record is defensible is positioning. Standard GPS scatters a few meters, which is enough to blur real change at a meadow edge. Adding RTK positioning and a heading source, on the order of one to two thousand dollars per boat, is what turns nice pictures into trustworthy year-on-year change detection.
The toolkit is currently a single-file concept tool with two linked surfaces, built to be handed to a stronger code build later with real geographic assets.
The survey planner sizes one survey. It takes a platform, a detail preset, a speed, a sonar range, an overlap, and an area, and returns the swath, the line spacing, the number of lines, the distance, the time in the water, the coverage rate, and an honest read on the smallest meadow patch the survey could reliably outline. It draws the track over the real coast so the density of the survey is visible, not just tabulated.
The fleet constructor turns an area and a budget into a proposal. It walks from tool choice, through sizing a fleet against the area with a live deployment map, to a delivery model, to the cost charts, to the people and capabilities behind it. It exists to answer the question a mayor actually asks, which is not how a sonar works but what it would take to have this and keep it.
The two surfaces share the same survey math. In the next build that math should be pulled into one module so the planner, the constructor, and any future routing or site-assessment view all call the same engine.
The same capability can reach a community three ways, and the right one depends on how committed the community is to running its own record.
Renting it is the one-time survey. Island Lab mobilizes the kit and crew, runs the survey to a fixed protocol, and hands over the finished map and the raw data. There is no capital and no upkeep. This is the path to a first baseline, a one-off assessment, or a test of the idea before anyone commits.
Owning it is the community fleet. The community buys the fleet outright and runs the survey itself each year, with Island Lab supplying the protocol, training the operators, and optionally staying on for quality control. This is the path to a truly sovereign, self-run record, and it is the one to hand to a fishermen's cooperative, a port authority, or municipal environment staff. It carries real upkeep, and the design says so plainly: batteries, connectors, transducer care, software updates, annual recalibration, and a refresh of operator training. Protocol drift is the risk to manage. Same lines, same settings, same season, or the record stops being comparable.
Running it as a program is the annual managed event. Island Lab owns and maintains the kit and brings it each season, the community paddles the lines as a shared event, and Island Lab delivers the mosaic and a change report against the previous year. It gives a community an ongoing record and genuine participation without the burden of owning and maintaining hardware.
A survey is a small team, not a piece of equipment. It needs paddlers or pilots to run the lines to the protocol, roughly one per boat and fewer if the platform is autonomous. It needs a survey lead who owns the plan, the lines, and the quality on the day. It needs a data person to turn the logs into the georeferenced mosaic and the outlines. And it needs a ground-truth capability, a diver or a drop-camera, to confirm what the acoustic textures actually are. In the rent and program models Island Lab supplies most of these. In the own model the community grows into them, which is the whole point of the hand-off.
The toolkit is built in the Island Lab house style so that everything the community sees is consistent and legible.
Typography pairs Fraunces for headings, Inter for body, and IBM Plex Mono for data and provenance. The palette runs on an abyss background with signal cyan for computed output, a slate blue for the assumptions a user sets, and a kelp gold for illustrative placeholders. Every visible number wears a provenance pill so its status is never in doubt. The underlying frame is the Three Truths: the global model, the local sensor, and the reconciled signal that comes from putting them together. Maps and demos are single-file and self-contained so they can be handed on and rebuilt cleanly.
Side-scan maps extent, not meaning, so ground-truth points are required, not optional. The swath collapses in the very shallowest fringe of the coast, where the geometry runs out, so the shallowest water is always the least certain. Sparse or patchy meadow edges stay ambiguous until a spot check resolves them. Standard positioning is fine for a first look but not for tight year-on-year overlays, which is why RTK is the headline upgrade. And a community-owned survey is only as good as the protocol it follows, which is why the protocol, not the sonar, is the thing we actually deliver.
None of these are reasons not to do this. They are the reasons to do it carefully, and stating them plainly is part of why a community can trust the result.
The immediate work is to replace the placeholder economics with a real cost model, so the labor rates, upkeep percentages, margins, and setup fees are defensible in front of a partner or a grant reviewer. The survey engine should be refactored into a single shared module that every surface calls. The schematic coast and area have now been replaced with the real meadow polygon and a real base layer, so the maps show Pollica rather than an illustration. And the whole capability should be positioned within the larger toolbox as one piece of a site assessment, routing, and planning offering, so a mayor sees not just a survey but a way to understand and act on the coast in front of them.
The larger ambition is a replicable template. A method and a toolkit that Pollica can run, that the next emblematic community can adopt, and that leaves each of them holding their own record of their own coast, year after year.
Data layer: Posidonia oceanica extent from EMODnet Seabed Habitats (CC BY 4.0), source data ISPRA, 2017 cartography.