On-demand drift modelling for search and rescue
SEASCOPE-DRIFT documentation
User guide and system description
1 Introduction#
SEASCOPE-DRIFT is an on-demand drift modelling system for marine search and rescue (SAR) and, more generally, for tracking objects and material at sea. It predicts where persons, life rafts, vessels and debris drift from a last known position, and turns that prediction into a search plan. It was developed by Calypso Science and Oceanum, in collaboration with New Zealand Police and Coastguard New Zealand, around the needs of operational SAR teams.
This guide follows the application as it is: the screenshots are taken from the version named in the header, and the ranges and defaults of every control are read from that version at each update.
| Component | Version documented | Release date |
|---|---|---|
| Drift engine | v0.3.1 | 2026-10-22 |
| Drift-UI | v0.1.15 | 2026-10-22 |
1.1 Purpose and intended users#
SEASCOPE-DRIFT is used to simulate drift trajectories of objects or persons at sea and in coastal waters, and to plan the resulting search. Its users are:
- SAR coordinators, who set up the incident, run forward or backward drift predictions, compare wind forecasts and design search patterns;
- operational staff (police, coastguard, air units), who receive search areas and patterns as maps and GPS routes.
1.2 System overview#
SEASCOPE-DRIFT is a web application, used from any modern browser with nothing to install. The simulations run on Oceanum's cloud platform, on a drift engine fed with the best ocean and wind model data available for each region — high-resolution coastal models where they exist, relaxing to GFS or ECMWF winds and Mercator SMOC surface currents elsewhere (Section 5). A 48-hour simulation of a few thousand particles typically returns in seconds.
The application computes no physics in the browser: every current, particle position and tide curve comes from the drift engine, so what the screen shows is exactly what was simulated. Each region has its own address, its own models and its own defaults (Section 5.3).
1.3 About this documentation#
This documentation is regenerated as the application evolves: at each update the screenshots are re-taken from the current release, the tables of controls are re-read from it, and what changed is listed on the What's new page.
2 Getting started#
2.1 Opening the application#
The Drift-UI opens in a web browser at the address of the region's deployment; nothing is installed on the user's computer. Each region has its own address, and the deployment decides who may open it (an Oceanum login, or open access). At start-up the quick guide summarises every control and the usual workflow: set a datum, run, compare, plan, export. Tick Don't show this again to skip it next time; it can be re-opened from the About box.

2.2 A typical incident, end to end#
- Check the forcing. Choose Dynamic mode for forecast currents and wind (or Tide mode for a short, sheltered-water case) and open the domain pill to check that the forecast covers the period of interest (Section 3.3).
- Set the datum. Right-click the map at the last known position and choose New datum, or press Edit position in the Release panel and type the coordinates. Set the seed radius to the uncertainty of the position, and the release window to the uncertainty of the time (Section 3.4).
- Set the time. Enter the time of the incident, the duration, and the direction — forward to predict where the object is now, backward to find where an object found at sea came from (Section 3.5).
- Choose the wind and the object. Choose the forecast wind (ECMWF or GFS) and the drifting object — usually a Leeway category such as a person in the water with a buoyancy aid, or a life raft (Section 3.6).
- Run. Press the red Run button. The result appears within seconds and plays through the drift; drag along the timeline to the time the search units will be on scene (Section 3.7).
- Compare. Save the run, switch to the other wind forecast and run again: the two ensembles side by side show how much the answer depends on the wind forecast (Section 3.8).
- Plan. Open the search pattern toolbox: a search area is proposed from the particle cloud; adjust it, choose the pattern and track spacing, name it and save it (Section 3.9).
- Export and share. Export the patterns as GPX or KML for the units, and the drift as KML, GeoJSON or CSV for the GIS (Section 3.11). Re-run as new information arrives.
SEASCOPE-DRIFT supports, and does not replace, the judgement of the SAR mission coordinator. A drift prediction is a probability distribution, not a position: plan on the spread of the ensemble, not on its centre alone (Section 7).
3 Drift-UI user guide#
This chapter describes the Drift-UI screen by screen. Each screenshot is followed by a table of its numbered controls: what each does, and its range and default as the code of the documented version has them. Some defaults — the release position, the time step, the seed radius, the starting mode — are set per region (Section 3.14); the screenshots are from the Drift-UI v0.1.15 in the Hauraki Gulf region.
3.1 Screen layout#
The map fills the screen. The three pills at the top centre identify the run — the application, the simulation mode and the domain. The five round tool buttons at the top right open the panels. The colour scale and the cursor read-out sit at the top left, the map tools at the bottom left, the timeline at the bottom centre, and the run buttons at the bottom right. Right-clicking the map opens a context menu (Section 3.4).

| No. | Control | What it does | Range / default |
|---|---|---|---|
| 1 | SEASCOPE-DRIFT pill | Opens the About box, which also re-opens the quick guide. | |
| 2 | Mode pill | Shows the simulation mode, Tide or Dynamic; opens the mode panel. | |
| 3 | Domain pill | Shows the domain in use — the tidal dataset in tide mode, the forecast stack in dynamic mode; opens the domain panel. | |
| 4 | Release | Opens the Release panel — where, how many particles, how spread out. | |
| 5 | Time | Opens the Time panel — start, duration, step, forward or backward. | |
| 6 | Wind & object | Opens the Wind and drifting object panel. | |
| 7 | Search pattern | Opens the search pattern toolbox. | |
| 8 | Layers & display | Opens the layers panel, the saved runs and patterns, and export. | |
| 9 | Colour scale | Scale of the current speed (and the wind and density, when those layers are on). | |
| 10 | Timeline | Play and step buttons, the clock (local time and UTC), the tide gauge and its sea level, and the tide curve or time slider. | |
| 11 | Measure | Click points on the map for the distance (nautical miles) and true bearing of each leg, and the total; double-click to finish, Esc or the button again to clear. | |
| 12 | Zoom in | Zooms in (also mouse wheel). | |
| 13 | Fit domain | Zooms back to the whole region. | |
| 14 | Save run | Keeps the finished run as a layer in its own colour, after asking for a name, so another run can be compared with it. | |
| 15 | Run | Starts the simulation. A progress card shows the elapsed time; the result appears within seconds and starts playing. Esc or ✕ cancels the wait. |
Only one panel is open at a time; pressing its tool button again, pressing Esc or clicking the map closes it (except the search pattern and release panels while they are being used on the map). In dynamic mode the map animates the forecast surface currents instead of the tide:

3.2 About and quick guide#
The SEASCOPE-DRIFT pill opens the About box: what the system is, who built it, and its disclaimer. Its button Show the quick guide re-opens the guide shown at start-up (Section 2.1).

3.3 Simulation mode and domain#
The mode pill opens the mode panel. Tide mode drives the drift with tidal currents reconstructed from harmonic constituents, which are available at any date, past or future, and a uniform wind entered by the user. Dynamic mode drives it with the forecast current stack and a forecast wind (Section 4.8). The mode the page opens in is set per region. In a region with no tide model, Tide is greyed out and the page opens in Dynamic; Dynamic is greyed out while no forecast is staged for the domain.

| No. | Control | What it does | Range / default |
|---|---|---|---|
| 1 | Simulation mode | Tide drives the drift with tidal currents reconstructed from harmonic constituents, with a wind you enter. Dynamic uses the forecast current stack and a forecast wind. Custom is reserved and disabled. | Tide · Dynamic |
| 2 | Mode description | What the selected mode uses and when to choose it. |
The domain pill opens the domain panel. In tide mode it selects the tidal dataset:

| No. | Control | What it does | Range / default |
|---|---|---|---|
| 1 | Tidal dataset | The tidal model driving tide mode, among those available in the region. | |
| 2 | Dataset description | The tidal model, its resolution and constituents. |
In dynamic mode there is nothing to choose — the order of the ocean models is fixed for the region — so the panel lists the forecast stack instead: each current and wind source in priority order, the period it covers and when it was staged. Check it before a long run: the sources do not all reach the same horizon.

| No. | Control | What it does | Range / default |
|---|---|---|---|
| 1 | Forecast stack | Each current and wind source in priority order, with its type, time step, the period it covers and when it was staged. | |
| 2 | Stack note | How the stack is combined (the finest source answers wherever it covers a particle). |
3.4 Release#
The Release panel sets the last known position (the datum) and the ensemble of particles released there.

| No. | Control | What it does | Range / default |
|---|---|---|---|
| 1 | Coordinate format | How positions are written and typed — decimal degrees (DEG), degrees and decimal minutes (DMD) or degrees, minutes and seconds (DMS). The choice is remembered by the browser and also used by Copy location. | 12.3456° · 12° 34.56' · 12° 34' 56" |
| 2 | Latitude | Latitude of the release point (datum), in the chosen format. Locked until Edit position is pressed. | default set per region |
| 3 | Longitude | Longitude of the release point. | default set per region |
| 4 | Edit position | Unlocks the position. The cursor becomes a crosshair; a click on the map or a typed coordinate moves the point. Save locks it again, Cancel (or Esc) restores the previous position. The lock stops a stray click from moving an agreed datum. | |
| 5 | Position check | Whether the point is in the domain, the current there and which source provides it — or a warning that the point is on land. | |
| 6 | Particles | Size of the ensemble. More particles give a smoother density map, at a modest cost in run time. | 1–10,000; default 1,500 |
| 7 | Seed radius | Particles are seeded uniformly over a circle of this radius around the release point — the uncertainty of the last known position. | 0–2,000 m; default 100 m, or the region's |
| 8 | Release over | Spreads the release evenly over this many minutes after the start time, for an uncertain time of loss. 0 is an instantaneous release. | 0–360 min; default 0 min |
| 9 | Diffusivity | Horizontal diffusivity K of the random walk that represents unresolved current variability (standard deviation √(2 K Δt) per step and component). | 0–50 m²/s; default 0.2 m²/s |
The position is locked once saved, so that a stray click on the map cannot move an agreed datum. To move it, press Edit position — the cursor becomes a crosshair, and a click on the map or a typed coordinate moves the point (Enter in a coordinate box saves) — then Save to lock it again, or Cancel / Esc to put it back. The note under the buttons checks the point: in the domain, the current there and which model provides it, or a warning that it is on land. A release outside the region is refused.
The quickest way to start is the right-click menu on the map:

| No. | Control | What it does | Range / default |
|---|---|---|---|
| 1 | New datum | Places the release point here and walks through Time and Wind & object (each step can be skipped). Run at any point keeps the position. | |
| 2 | New search pattern | Opens the search pattern toolbox with the area centred here instead of on the drift. | |
| 3 | Copy location | Copies the position to the clipboard, in the chosen coordinate format. |
New datum opens a short guided set-up: the position is placed where you clicked, then the Time and the Wind & object panels follow (Step 2 of 3, Step 3 of 3), each with a Skip button — their defaults already run. Pressing Run at any point keeps the position and runs.
Regions that define named release points (offshore installations, for instance) offer them in a Preset location list under the coordinates, and draw them on the map.
3.5 Time#

| No. | Control | What it does | Range / default |
|---|---|---|---|
| 1 | Start time | Start of the simulation (for a backward run, the time the object was found). In regions with a time zone it is entered in local time and the UTC equivalent is shown below it; otherwise in UTC. In dynamic mode it cannot be earlier than the data — an earlier start is moved to the first available time. | |
| 2 | UTC equivalent | The start time in UTC, when the start is entered in local time. | |
| 3 | Duration | Length of the simulation. | 0.5–96 h; default 48 h |
| 4 | Step | Integration time step. Shorter steps follow sharp current gradients more closely and cost more. | 5–900 s; default 300 s or the region's, whichever is shorter |
| 5 | Direction | Forward predicts where the object goes; Backward estimates where an object found at sea came from. | Forward · Backward |
Times are entered in local time in regions that define a time zone, with the UTC equivalent shown underneath; otherwise in UTC. In dynamic mode the start cannot be earlier than the data: a start typed or reached earlier is moved to the first available time, with a note saying so. The data can reach back up to about a week, kept from past forecasts (Section 5.5), and forward to the end of the forecast; the domain panel shows the period covered.
A backward run starts at the time and place an object was found and integrates back in time, giving the ensemble of positions it may have come from (Figure 17).
The timeline at the bottom shows the clock and, where a sea level is available, the tide curve at a gauge. Before a run, dragging along it (or pressing play) moves the clock through the available period and animates the currents; after a run it scrubs through the result.

| No. | Control | What it does | Range / default |
|---|---|---|---|
| 1 | Step back | Moves the clock one step earlier (10 minutes, or the model output step if longer). | |
| 2 | Play / pause | Plays the run through its output frames. Before a run it sweeps the clock through the available period, animating the currents. | |
| 3 | Step forward | Moves the clock one step later. | |
| 4 | Clock | The displayed time — local time where the region has a time zone, with UTC below. | |
| 5 | Tide gauge | The gauge whose sea level is drawn as the curve. Tide gauges of the tidal model in tide mode; in dynamic mode, the region's gauges read from the forecast sea level (where configured). | |
| 6 | Sea level | Sea level at the selected gauge at the displayed time. | |
| 7 | Fold tide curve | Hides or shows the tide curve, leaving the play buttons and clock. |
3.6 Wind and drifting object#
This panel sets the wind and what it acts on. In tide mode the wind is uniform and steady, entered as a speed in knots and the direction it blows from:

| No. | Control | What it does | Range / default |
|---|---|---|---|
| 1 | Wind on/off | With the wind off only the current moves the particles, and the object settings do not apply. Off at start in tide mode; switching to dynamic mode turns it on. | off at start |
| 2 | Wind speed | Tide mode only — a uniform, steady wind speed. | 0–80 kt; default 15 kt |
| 3 | Wind from | Tide mode only — the direction the wind blows from, in degrees true (meteorological convention). The compass rose below shows it. | 0–360 °; default 270 ° |
| 4 | Wind rose | The wind direction drawn on a compass. | |
| 5 | Object type | What the wind acts on: a windage (a fixed percentage of the wind), a Leeway object from the US Coast Guard table, or surface oil (a 3 % windage by default, no weathering or spreading). | Windage % · Leeway object · Surface oil |
| 6 | Windage | Percentage of the 10 m wind added to the current. | 0–6 %; default 3 % |
| 7 | Deflection | Rotates the wind-driven drift clockwise (positive) or anticlockwise from downwind. | -40–40 °; default 0 ° |
In dynamic mode the wind is a forecast. The panel offers the global models available (ECMWF and GFS) and lists, for the chosen one, each wind source that will be used — including 1 km regional nests where the region has them — with its area, grid spacing and period:

| No. | Control | What it does | Range / default |
|---|---|---|---|
| 1 | Wind on/off | With the wind off only the current moves the particles, and the object settings do not apply. Off at start in tide mode; switching to dynamic mode turns it on. | off at start |
| 2 | Forecast wind | Dynamic mode only — which global wind forecast drives the run (ECMWF or GFS). Where the region has 1 km nests (e.g. PredictWind) they are used inside their box, above the chosen model. | |
| 3 | Wind coverage | The period the selected wind forecast covers, and its time step. | |
| 4 | Object type | What the wind acts on: a windage (a fixed percentage of the wind), a Leeway object from the US Coast Guard table, or surface oil (a 3 % windage by default, no weathering or spreading). | Windage % · Leeway object · Surface oil |
| 5 | Leeway object | One of the US Coast Guard Leeway categories (persons in water, life rafts, vessels, debris…), grouped. | |
| 6 | Leeway coefficients | The object's downwind and crosswind leeway as a percentage of the wind. | |
| 7 | Jibe / hour | Probability per hour that an object flips to the other side of downwind. Converted to a per-step rate, so the result does not depend on the time step. | 0–0.5 /h; default 0.04 /h |
| 8 | Downwind leeway | Applies the downwind component of the leeway. Untick to see the crosswind effect alone. | on at start |
| 9 | Crosswind leeway | Applies the crosswind component, which produces the two lobes of a Leeway ensemble. Untick to see the downwind drift alone. | on at start |
| 10 | Leeway at this wind | What the object does at a given wind speed — downwind and crosswind speeds and the divergence angle of its two lobes. |
The drifting object is one of:
- Windage % — a fixed percentage of the 10 m wind added to the current, optionally deflected; for objects without a Leeway category (Section 4.3);
- Leeway object — one of the 85 US Coast Guard Leeway categories (Section 4.2). The panel shows its coefficients and what it does at a given wind speed. The jibe rate and either component can be changed, for example to see the crosswind effect alone;
- Surface oil — a windage drift of its own, with no deflection and no weathering or spreading:
| Control | What it does | Range / default |
|---|---|---|
| Oil windage | Surface oil only — the windage applied to the slick. | 0–6 %; default 3 % |
With the wind switched off only the current moves the particles, and none of the object settings apply.
3.7 Running a simulation and reading the result#
The red Run button starts the simulation. A small progress card shows what is being done and the elapsed time; the map stays visible underneath. A run usually takes a few seconds; Esc or the ✕ button stops waiting for it. Running an identical simulation again — same release, settings and forcing — returns the stored result at once (Section 4.7).
When the result arrives it plays through the drift. The particles are drawn with their tracks, the run statistics card gives the state of the ensemble at the displayed frame, and the cursor read-out gives the position and current under the mouse:

| No. | Control | What it does | Range / default |
|---|---|---|---|
| 1 | Run statistics | At the displayed frame — integration step and frame, particles adrift and beached, how many each current source drives, the ensemble centroid, the 95 % radius and the time. | |
| 2 | Colour scale | Scale of the current speed (and the wind and density, when those layers are on). | |
| 3 | Cursor read-out | Position under the cursor and the current there (speed and direction it flows to). | |
| 4 | Timeline | Play and step buttons, the clock (local time and UTC), the tide gauge and its sea level, and the tide curve or time slider. | |
| 5 | Run details | Opens the run details — the settings the run was made from and its result. The same details head every exported file. | |
| 6 | Save run | Keeps the finished run as a layer in its own colour, after asking for a name, so another run can be compared with it. | |
| 7 | Clear / cancel | Clears the run from the map. While a run is computing, cancels it. |
The statistics card lists, at the displayed time, how many particles are adrift and how many have beached, how many each current model is driving (so you can see when the drift leaves a high-resolution domain), the ensemble centroid and the 95 % radius — the distance from the centroid within which 95 % of the particles lie. If a run is refused — outside the forecast period, on land, outside the region — the card says why and, where it can, what to change.
The same run as a density map shows the probability of presence at the displayed time (Layers panel, Density):

The ⓘ button opens the run details: everything the run was made from (release, time, forcing sources and when they were staged, object and its coefficients) and what it came to. The same rows head every file the run is exported to, so an exported file is self-describing.

A backward run (Time panel, Backward) is read the same way: the cloud is where the object may have been at the displayed time before it was found.

3.8 Saved runs and comparison#
The save button (the floppy disk above Run) keeps the finished run as a layer in its own colour, under a name you give it — by default its object, wind and start time.

| No. | Control | What it does | Range / default |
|---|---|---|---|
| 1 | Run name | Name of the saved run, pre-filled from its object, wind and start time. It names the layer and the exported files. |
Saving a run, changing one setting and running again puts the two on one chart. Comparing the ECMWF and GFS wind forecasts this way is the most direct way to see the forecast uncertainty; the same works for two Leeway objects, two release times or two seed radii.

Saved runs are listed in the Layers panel, where each can be shown or hidden, opened in run details (ⓘ), downloaded (↓) or deleted (×) (Figure 24). They are kept by the page for the session: reloading the page clears them, so export what must be kept.
3.9 Search patterns#
The search pattern toolbox designs a search over the drift. When it opens after a run, a search area is proposed from the particle cloud at the displayed time — fitted on the cloud's principal axis — and follows the cloud as the timeline moves, until you take it over. It can then be moved (drag the middle), resized (drag an edge) and rotated (drag a corner) on the map, drawn afresh (Draw area), or reset to the cloud. New search pattern in the right-click menu starts one at the clicked point instead.

| No. | Control | What it does | Range / default |
|---|---|---|---|
| 1 | Pattern | Parallel track, expanding square or sector search. Geometry after the US Coast Guard SAROPS formulation (Kratzke, Stone & Frost, 2010). | Parallel · Expanding square · Sector |
| 2 | Draw area | Drag a new search area on the map, replacing the proposed one. | |
| 3 | Reset to cloud | Goes back to the area fitted automatically to the particle cloud, which follows the timeline. | |
| 4 | Search area | Size and orientation of the search area, and how to change it on the map (the middle moves, edges resize, corners rotate). | |
| 5 | Track spacing | Distance between parallel legs (or between successive legs of the expanding square). | 100–4,000 m; default 500 m |
| 6 | Along | Length of the search area along the legs (parallel track). | 500–30,000 m; default 6,000 m |
| 7 | Across | Width of the search area across the legs (parallel track). | 500–30,000 m; default 4,000 m |
| 8 | Orientation / first leg | Orientation of the legs (parallel track) or direction of the first leg (square, sector), in degrees true. | 0–359 °; default 0 ° |
| 9 | First turn | Which way the craft turns at the end of the first leg, and therefore at every leg after it. | Right · Left |
| 10 | Name | The pattern's name. It follows the pattern into the saved layer, the file name and the route name inside GPX and KML files. | |
| 11 | Save pattern | Keeps the pattern as a layer, listed under Search patterns in the layers panel, where it can be shown, hidden and exported. | |
| 12 | Pattern summary | Number of legs, distance to fly (km and nautical miles), spacing and area. |
Three patterns are available, with their geometry after the US Coast Guard SAROPS formulation (Kratzke, Stone & Frost, 2010):
- Parallel track — parallel legs over a rectangle, with the track spacing, length along, width across, orientation and first turn;
- Expanding square — legs of increasing length outwards from the datum, at the track spacing, starting in the chosen direction;
- Sector — legs through the datum at fixed angles, over a radius, repeated for a number of rounds:
| Control | What it does | Range / default |
|---|---|---|
| Search radius | Radius of the expanding square or sector search around the datum. | 200–20,000 m; default 2,000 m |
| Rounds | Sector search — number of rounds (each round is a set of legs through the datum, rotated from the previous one). | 1–6; default 3 |



The summary under the controls gives the number of legs and the distance to fly. A green dot marks the start and a red dot the end. Save pattern keeps it as a named layer; saved patterns are listed in the Layers panel and are what Search patterns (shown) exports (Section 3.11).
3.10 Layers and display#
The layers panel controls what is drawn on the map, lists the saved runs and patterns, and holds the export.

| No. | Control | What it does | Range / default |
|---|---|---|---|
| 1 | Satellite imagery | Satellite basemap under the map. | on at start |
| 2 | ENC | Nautical chart overlay (in New Zealand, LINZ charts). | off at start |
| 3 | Current | Animated current particles, coloured by speed (scale in the legend). | on at start |
| 4 | Wind | Animated wind layer, from the selected forecast (dynamic mode). | off at start |
| 5 | Model domains | Outlines of the model domains in the current stack, so you can see where each model drives the drift. | off at start |
| 6 | Tracks | Particle tracks behind the particles. | on at start |
| 7 | Density | Particle density map — the probability of presence at the displayed time. | off at start |
| 8 | Density cell | Size of the density map cells, in screen pixels. | 4–28 px; default 10 px |
| 9 | Track colour | Colours tracks by particle age, by speed, or in a single colour. | by age · by speed · single colour |
| 10 | Saved runs | Each saved run in its colour, with a checkbox to show or hide it, ⓘ for its run details, ↓ to download it and × to delete it. | |
| 11 | Saved search patterns | Each saved pattern — tick to show it (and include it in a pattern export), ↓ to download it, × to delete it. | |
| 12 | Export — what | The run on screen, any saved run, all saved runs (one file each, in a zip), or the search patterns that are ticked (shown). | |
| 13 | Export — format | The formats available for the selected item (see the export table). | |
| 14 | Centreline only | Drift tracks only — export the path of the ensemble centroid alone, without the particles. A full export includes the centreline too. | off at start |
| 15 | Export | Writes and downloads the file. Red when there is something to export. |



3.11 Export#
The export section at the bottom of the layers panel writes the drift or the search patterns to a file. The formats on offer follow what is being exported:
| What | Formats |
|---|---|
| Drift tracks (run on screen, or a saved run) | CSV, GeoJSON, GPX, KML |
| All saved runs (zip, one file each) | CSV, GeoJSON, GPX, KML |
| Search patterns (those shown) | GeoJSON, CSV, GPX, KML, GeoPackage |
What the files contain:
- Drift tracks — KML: one time-stamped track per particle, which Google Earth both draws and plays on its time slider, the centreline (path of the ensemble centroid) drawn heavier, and the release point. Thinned to keep the file loadable: up to 500 particles and about 120 instants, always including the last frame. The run details are attached as extended data.
- Drift tracks — GPX: the centreline and the particle tracks as GPX tracks with times, thinned as for KML.
- Drift tracks — CSV: one row per particle per output frame (
particle, time, lon, lat, status), every particle, then the centreline; the run details are written as#comment lines at the top. - Drift tracks — GeoJSON: a time-stamped LineString per particle with its final status, and the centreline, with the run details as properties and as collection metadata.
- Centreline only: the centroid path alone, in any of these formats.
- All saved runs: every saved run as its own file, in one zip.
- Search patterns: each shown pattern as a route — GPX
<rte>, KML LineString with start and end placemarks, GeoJSON LineString carrying the pattern parameters, CSV of turn points, or GeoPackage (for GIS).
Particle status codes are 0 adrift, 1 beached, 2 out of the domain and 3 not yet released. A saved run is exported from the copy the page holds, so it can be exported for as long as the page is open.
3.12 Measuring distances#
The ruler button at the bottom left measures on the chart: click points for the distance in nautical miles and the true bearing of each leg, and the total; double-click to finish; Esc or the button again clears it.

3.13 Keyboard and mouse#
| Action | Effect |
|---|---|
| Drag the map | Pan |
| Mouse wheel, + / − | Zoom; ⤢ fits the region |
| Right-click the map | Context menu: New datum, New search pattern, Copy location. While a position is being edited, a right-click places it |
| Esc | Closes the innermost thing open: cancels a run being computed, then a dialog, the menu, a measurement, an area being drawn, a position being edited, and finally the open panel |
| Enter in a coordinate box | Saves the position being edited |
| Drag along the timeline | Moves the clock; step buttons move one step |
3.14 Regional settings#
The same application serves every region; what differs is set per region:
- the title and subtitle, the map frame, the attribution line and the About text;
- the time zone used for entering and showing times;
- the default mode, wind source, release point, windage, time step and seed radius;
- tide gauges read from the forecast sea level, for regions without a tide model;
- preset release points, offered in the Release panel and drawn on the map;
- whether the model domains layer is on at start;
- partner logos under the tool buttons;
- the interface language (en, fr available).
The controls are capped at the limits of the region (maximum particles, duration, minimum time step).
4 Drift model#
This chapter explains what a SEASCOPE-DRIFT simulation computes, so that its results can be read with the right confidence.
4.1 Lagrangian particle tracking#
The search object is represented by an ensemble of particles — up to 10,000 per run. At each time step every particle moves with the surface ocean current at its position plus a wind-driven drift that depends on the object (Section 4.2, Section 4.3). Currents and winds are interpolated in space and time from the model data of the region (Section 5). Simulations last up to 96 hours, with a time step of 5 to 900 seconds (Time panel, Section 3.5).
4.2 Wind-driven drift: the Leeway model#
SAR objects are driven by the US Coast Guard Leeway model (Allen and Plourde, 1999; Allen, 2005), as implemented in OpenDrift (Dagestad et al., 2018) and on which the IAMSAR leeway tables are based. The leeway of an object is split into a downwind and a crosswind component, each a linear function of the 10 m wind speed W (m/s):
The slopes and offsets are the object's coefficients from the US Coast Guard table (Section 9). The random terms ε are drawn once per particle at release from the published standard deviations — this is what spreads a Leeway ensemble even in a uniform wind — and redrawn where they would send an object upwind. Each object settles on one side of downwind (half the ensemble on each side) and keeps it until it jibes; the jibe probability is set per hour (default 0.04 per hour) and does not depend on the time step. This produces the two divergent lobes of a Leeway ensemble that SAR planners expect. Either component can be switched off to see the other alone.
4.3 Generic windage and surface oil#
For objects without a Leeway category, a windage adds a fixed percentage of the 10 m wind to the current, optionally deflected from downwind. Surface oil uses the same drift at 3 % of the wind by default, with no deflection; no weathering, evaporation or spreading of the oil is modelled.
4.4 Spreading and uncertainty#
The prediction is stochastic: its uncertainty is the spread of the particle ensemble, which comes from four sources:
- the per-particle Leeway coefficients and jibing (Section 4.2);
- a horizontal random walk representing current variability the models do not resolve, set by the diffusivity K (Release panel; each step moves a particle by a random distance of standard deviation √(2 K Δt) in each direction);
- the seed radius: particles are spread uniformly over a circle around the last known position, representing its uncertainty;
- the release window: particles are released evenly over a period, representing the uncertainty of the time of loss.
From the ensemble come the density map, the 95 % radius and the search area proposed in the search pattern toolbox.
4.5 Forward and backward simulation#
A simulation runs forward, to predict where the object is now or will be, or backward, to estimate where an object found at sea came from. A backward run integrates the same equations back in time; the ensemble still widens going back, as the uncertainty about the origin grows.
4.6 Release and coastline#
Particles that would start on land are moved into the water; a release point on land is refused. A particle that reaches the coastline is stranded (beached) and stays there. In tide mode the coastline is that of the tidal model; in dynamic mode it is a detailed coastline prepared for the region (in New Zealand, LINZ 1:50,000). The coastline drawn on the map is the one the simulation uses, and the statistics count particles adrift and beached — beaching locations indicate the shorelines to search.
4.7 Reproducibility#
A simulation is reproducible: the same release, settings and forcing give the same result, and running it again returns the stored result at once instead of recomputing it.
4.8 Simulation modes#
Tide mode drives the drift with tidal currents reconstructed from the harmonic constituents of a tidal model (for the Hauraki Gulf: Z0, M2, S2, N2, K2, M4, MS4, MN4). The tide is computed analytically, so it is available at any date, past or future, independently of any forecast; the wind is uniform and entered by the user. Tide mode suits sheltered, tide-dominated waters, quick what-if analyses, and dates outside the forecast period.
Dynamic mode drives the drift with forecast ocean currents — which include the tide, the wind-driven and the large-scale circulation — and the forecast wind of the chosen global model, each from the best data available along the path (Section 5). It is recommended when winds are strong or changing, offshore, and for drifts longer than a few hours. Running the same incident with both wind forecasts shows the forecast uncertainty directly (Section 3.8).
5 Best available ocean and wind data#
Every SEASCOPE-DRIFT simulation is driven by the best model data available along the path of each particle. Where Calypso Science runs a high-resolution hydrodynamic model — typically resolving harbours, estuaries and headlands at tens to a few hundred metres — that model drives the drift. Where a 1 km PredictWind downscale of the wind exists, it drives the wind-driven part of the drift. Outside these domains the simulation relaxes to global models:
- wind: the 10 m wind of the GFS (NOAA) or ECMWF global forecast, the one chosen in the Wind & object panel (Section 3.6);
- ocean currents: the Mercator Ocean SMOC global surface currents (Copernicus Marine), which combine the large-scale circulation, the tide and the wave-induced drift, at 1/12° (about 8 km), hourly.
The switch happens particle by particle and step by step, so a drift that starts in a harbour and leaves for open water is driven by the harbour model, then by the coarser models as it goes. The run statistics show how many particles each model is driving at the displayed time (Section 3.7).
5.1 High-resolution domains#
The map shows the high-resolution domains currently used around New Zealand and in France. Extents are read from the model metadata at each documentation update.
| Area | Domain | Type | Extent (W, S, E, N) |
|---|---|---|---|
| New Zealand | SCHISM Hauraki | High-resolution ocean forecast | 174.31, -37.22, 175.93, -35.62 |
| New Zealand | SCHISM NZ | National ocean model (SCHISM NZ) | 165.05, -49.15, 179.90, -33.14 |
| New Zealand | PredictWind 1 km — Great Barrier | PredictWind 1 km wind | 175.01, -36.67, 175.81, -35.87 |
| New Zealand | PredictWind 1 km — Tauranga | PredictWind 1 km wind | 175.80, -37.99, 176.60, -37.19 |
| New Zealand | Hauraki Gulf — SCHISM | Tidal model | 174.31, -37.22, 175.93, -35.62 |
| New Zealand | SCHISM Raglan | High-resolution ocean forecast | 174.70, -37.89, 174.97, -37.71 |
| New Zealand | PredictWind 1 km — Raglan | PredictWind 1 km wind | 174.27, -38.34, 175.07, -37.54 |
| New Zealand | Raglan Harbour — Oceanum tide | Tidal model | 174.75, -37.84, 174.98, -37.74 |
| New Zealand | SCHISM Wellington | High-resolution ocean forecast | 174.70, -41.43, 174.91, -41.23 |
| New Zealand | PredictWind 1 km — Wellington | PredictWind 1 km wind | 174.43, -41.64, 175.23, -40.84 |
| New Zealand | PredictWind 1 km — Picton | PredictWind 1 km wind | 173.77, -41.73, 174.57, -40.93 |
| New Zealand | Cook Strait — SCHISM | Tidal model | 174.03, -42.26, 175.40, -40.70 |
| New Zealand | SCHISM Nelson | High-resolution ocean forecast | 173.07, -41.33, 173.33, -41.17 |
| New Zealand | PredictWind 1 km — Nelson | PredictWind 1 km wind | 172.85, -41.47, 173.65, -40.67 |
| France | Gulf of Saint-Malo SCHISM | Tidal model | -3.18, 48.50, -1.02, 50.03 |
| France | Brest & Iroise SCHISM | Tidal model | -5.64, 47.79, -4.18, 48.75 |
| France | Quiberon SCHISM | Tidal model | -3.66, 47.08, -2.31, 47.82 |
| France | South Brittany — 250 m grid | Tidal model | -4.23, 46.78, -1.96, 47.93 |
| France | Pertuis Charentais SCHISM | Tidal model | -2.77, 45.23, -0.69, 46.54 |
- High-resolution ocean forecasts (dynamic mode): Calypso SCHISM models of harbours and coastal waters, nested in the national SCHISM New Zealand model;
- Tidal models (tide mode): high-resolution tidal models, from which the tidal currents are reconstructed at any date;
- PredictWind 1 km winds: downscales of ECMWF and GFS over complex terrain, used above whichever global wind is chosen.
5.2 How the models are combined#
The ocean models form an ordered stack, finest first. For each particle at each time step the finest model whose domain contains the particle provides the current, and coarser models take over as the particle leaves it. Inside a fine model's domain, that model also decides where the land is, so a coarse model can never put current where the fine model knows there is land.
Winds work the same way: a PredictWind 1 km nest answers inside its own box and forecast period, and the chosen global model everywhere else. The nest follows the global model chosen — ECMWF runs use the ECMWF downscale, GFS runs the GFS one — so comparing the two wind forecasts compares the whole wind chain.
The order of the ocean models is fixed for each region; the choice of wind forecast is the user's. The domain panel lists the models of the region in priority order, with the period each covers (Section 3.3).
5.3 Regions#
The same application serves several regions, each with its own models:
| Region | Modes | Ocean models (finest first) | Tidal models | Wind |
|---|---|---|---|---|
| North Atlantic | tide + dynamic | — then Copernicus global surface | Gulf of Saint-Malo SCHISM Brest & Iroise SCHISM | ECMWF 10 m wind GFS 10 m wind |
| South Brittany | tide + dynamic | — | Quiberon SCHISM South Brittany — 250 m grid Brest & Iroise SCHISM Gulf of Saint-Malo SCHISM Pertuis Charentais SCHISM | ECMWF 10 m wind GFS 10 m wind |
| Côte d'Ivoire | dynamic | — then Copernicus surface globale | — | ECMWF vent à 10 m GFS vent à 10 m nested: WRF Abidjan 3 km vent à 10 m |
| Hauraki Gulf | tide + dynamic | SCHISM Hauraki SCHISM NZ then Copernicus global surface | Hauraki Gulf — SCHISM Hauraki Gulf — Oceanum 400 m | ECMWF 10 m wind GFS 10 m wind nested: PredictWind 1 km Great Barrier, PredictWind 1 km Tauranga |
| Nelson | dynamic | SCHISM Nelson SCHISM NZ | — | ECMWF 10 m wind GFS 10 m wind nested: PredictWind 1 km |
| Raglan | tide + dynamic | SCHISM Raglan then Copernicus global surface | Raglan Harbour — Oceanum tide | ECMWF 10 m wind GFS 10 m wind nested: PredictWind 1 km |
| Phuket & Phang Nga Bay | tide + dynamic | — | Phuket & Phang Nga Bay — Oceanum 400 m | ECMWF 10 m wind GFS 10 m wind |
| Western Australia | dynamic | SCHISM WA then Copernicus global surface | — | ECMWF 10 m wind GFS 10 m wind |
| Wellington & Cook Strait | tide + dynamic | SCHISM Wellington SCHISM NZ | Cook Strait — SCHISM | ECMWF 10 m wind GFS 10 m wind nested: PredictWind 1 km Picton, PredictWind 1 km Wellington |
5.4 Coastlines#
Particles strand on a coastline prepared for each region — in New Zealand from the LINZ 1:50,000 coastline, elsewhere from other coastline data — and in tide mode on the coastline of the tidal model. The coastline drawn on the map is the one the simulation uses.
5.5 Forecast updates#
Forecasts are refreshed automatically as each model publishes a new cycle — typically daily for the ocean models and twice daily for the global winds. Each update also keeps the past: a rolling archive of up to about a week is built from the first part (the nowcast) of each past forecast, the best estimate of that period any forecast made, so a dynamic run can start in the recent past. Gaps in the record are never filled in. The domain panel shows, for each model, the period covered and when it was last updated.
6 Validation and performance#
6.1 Validation against OpenDrift#
The engine is validated against OpenDrift (Dagestad et al., 2018), the open-source trajectory framework of the Norwegian Meteorological Institute, widely used for operational SAR. Both models are run on the same forcing files (in the Hauraki Gulf: SCHISM Hauraki, SCHISM NZ and Copernicus currents with ECMWF wind), the same landmask and the same configuration, in three steps from the simplest to the most uncertain, so that each step adds one source of difference:
- one particle, current only — interpolation, integration and geodesy;
- 3 % windage, 1,500 particles — adds the wind; diffusion is the only random term;
- Leeway objects, 1,500 particles each — adds the Leeway coefficients drawn per particle.
The results below are those of the documented version of the drift engine.
Step 1 — one particle, current only#
Release 175.0520 E, 36.3650 S at 2026-10-11T00:00:00Z, 48 h, 100 particles, jibe 0/h, diffusivity 0 m2/s, 600 s step, wind_ec wind.
| quantity | value |
|---|---|
| separation at +6 h | 7 m |
| separation at +12 h | 11 m |
| separation at +24 h | 17 m |
| separation at +48 h | 38 m |
| separation at +48 h (end) | 38 m |
| max separation | 38 m |
| path length travelled | 27.8 km |
| final separation as % of path | 0.14% |

With no random term the two trajectories should coincide; their small separation comes from the interpolation of the currents and from SEASCOPE-DRIFT's spherical geodesy against OpenDrift's ellipsoidal one.
Steps 2 and 3 — windage and Leeway ensembles#
Release 175.0520 E, 36.3650 S (radius 100 m) at 2026-09-26T00:00:00Z, 48 h, 1500 particles, jibe 0.04/h, diffusivity 0.2 m2/s, 600 s step, wind_ec wind.
| run | centroid drift | API vs OpenDrift centroid gap at +48 h | Monte-Carlo error (1 sigma) | overlap (halves) |
|---|---|---|---|---|
| OceanDrift, 3% windage | 68.6 km | 0.13 km (0.2%) | 0.01 km | 1.00 (1.00) |
| Leeway PIW-5, person in water, scuba suit | 29.5 km | 0.57 km (1.9%) | 0.35 km | 0.76 (0.66) |
| Leeway LIFE-RAFT-DB-10, deep ballast, general | 70.4 km | 0.50 km (0.7%) | 0.37 km | 0.75 (0.65) |
| Leeway LIFE-RAFT-SB-6, shallow ballast and canopy, general | 48.8 km | 0.25 km (0.5%) | 0.83 km | 0.65 (0.54) |
Because the two models draw their random numbers from different generators, particle i is not the same object in both runs: the comparison is between ensembles. The gap between centroids is read against the Monte Carlo error of a centroid (spread/√n), and the overlap of the final-position histograms against the overlap two halves of the same ensemble achieve.



For objects whose leeway is poorly constrained — a person in the water especially — the plume is wide and the relative gap between the two ensembles is largest, but it remains small against the spread of the plume itself: for such objects the search area is set by the object, not by the choice of drift engine.
6.2 Automated tests#
The engine is also checked by an automated test suite against independent references rather than against earlier runs of the same code: geodesic distances against pyproj, tidal reconstruction against the oceantide package, and the Leeway model against a transcription of OpenDrift's implementation.
6.3 Performance#
The engine is built so that a prediction returns in seconds. The forecast period a run needs is read once into memory and kept for the following runs, the mesh and tidal interpolation reuse precomputed structures, and the land test is folded into the last integration stage. The progress card shows the elapsed time of each run, and the run details give the compute time (Section 3.7).
Run time grows roughly linearly with the number of particles and the number of time steps; the limits per run are 10,000 particles and 96 hours.
7 Limitations and good practice#
7.1 Data availability#
- Dynamic mode needs a forecast covering the whole simulation. Forecasts reach a few days ahead and the sources do not share the same horizon; the domain panel shows each source's coverage — check it before setting a long duration.
- In dynamic mode a run can start in the past only as far back as the archive of past forecasts reaches (about a week by default); the Start picker stops there. Tide mode is available at any date.
- Outside the high-resolution domains, currents come from coarser models that do not resolve small bays, harbours or headlands. The statistics card shows which source is driving the particles.
7.2 Forecast and model uncertainty#
- A drift prediction is a probability distribution, not a position. Plan on the density map and the spread of the ensemble, not on the centroid alone.
- Wind forecasts differ: run both ECMWF and GFS and treat the difference as part of the uncertainty.
- Leeway coefficients are statistical fits to field experiments; the choice of object category matters more than most numerical settings. When the state of the object is unknown, use the unknown state or general categories, or run several.
- Tide mode ignores wind-driven and large-scale currents and uses a uniform, steady wind; it suits sheltered, tide-dominated waters and short durations.
- Stranding is decided against a coastline resolved to tens to a hundred metres: beaching locations indicate the shoreline to search, not an exact point.
- Surface oil is modelled as a windage drift only: no weathering, evaporation, emulsification or spreading.
7.3 Recommended use#
- Use at least 1,000 particles for planning; more particles give a smoother density map at a small cost in run time.
- Set the seed radius and the release window to the real uncertainty of the last known position and time.
- Re-run as soon as new information arrives — a sighting, an updated position, a new forecast cycle.
- Use backward runs to test hypotheses about the origin of an object found at sea.
- Export what must be kept: saved runs and patterns live in the browser page for the session only.
- SEASCOPE-DRIFT supports, and does not replace, the judgement of the SAR mission coordinator and the decisions of the authority coordinating the search.
8 References#
Allen, A. A. (2005). Leeway divergence. US Coast Guard Research and Development Center, Report CG-D-05-05.
Allen, A. A., & Plourde, J. V. (1999). Review of leeway: field experiments and implementation. US Coast Guard Research and Development Center, Report CG-D-08-99.
Breivik, Ø., & Allen, A. A. (2008). An operational search and rescue model for the Norwegian Sea and the North Sea. Journal of Marine Systems, 69(1–2), 99–113.
Dagestad, K. F., Röhrs, J., Breivik, Ø., & Ådlandsvik, B. (2018). OpenDrift v1.0: a generic framework for trajectory modelling. Geoscientific Model Development, 11(4), 1405–1420.
IMO/ICAO. International Aeronautical and Maritime Search and Rescue (IAMSAR) Manual, Volume II: Mission Co-ordination.
Kratzke, T. M., Stone, L. D., & Frost, J. R. (2010). Search and Rescue Optimal Planning System. 13th International Conference on Information Fusion (FUSION), IEEE.
Zhang, Y. J., Ye, F., Stanev, E. V., & Grashorn, S. (2016). Seamless cross-scale modeling with SCHISM. Ocean Modelling, 102, 64–81.
9 Appendix — Leeway object categories#
The 85 categories of the US Coast Guard Leeway table offered in the Wind & object panel, in the order of the table, with what each does at a 10 m/s wind (mean of the ensemble coefficients). Further categories, for example specific aircraft debris, can be added on request.
| Id | Key | Description | Downwind (% of wind) | Crosswind (% of wind) | Divergence (°) |
|---|---|---|---|---|---|
| 1 | PIW-1 | Person-in-water (PIW), unknown state (mean values) | 0.96 | 0.54 | 29.4 |
| 2 | PIW-2 | PIW, vertical PFD type III conscious | 0.48 | 0.15 | 17.4 |
| 3 | PIW-3 | PIW, sitting, PFD type I or II | 1.60 | 0.13 | 7.7 |
| 4 | PIW-4 | PIW, survival suit (face up) | 1.71 | 1.36 | 29.5 |
| 5 | PIW-5 | PIW, scuba suit (face up) | 0.63 | 0.31 | 26.2 |
| 6 | PIW-6 | PIW, deceased (face down) | 1.12 | 0.04 | 11.4 |
| 7 | LIFE-RAFT-DB-10 | Life raft, deep ballast (DB) system, general, unknown capacity and loading (mean values) | 3.52 | 0.62 | 5.6 |
| 8 | LIFE-RAFT-DB-11 | 4-14 person capacity, deep ballast system, canopy (average) | 3.50 | 0.78 | 7.2 |
| 9 | LIFE-RAFT-DB-12 | 4-14 person capacity, deep ballast system, no drogue | 3.75 | 0.78 | 6.8 |
| 10 | LIFE-RAFT-DB-13 | 4-14 person capacity, deep ballast system, canopy, no drogue, light loading | 3.75 | 1.00 | 7.6 |
| 11 | LIFE-RAFT-DB-14 | 4-14 person capacity, deep ballast system, no drogue, heavy loading | 3.59 | 0.48 | 7.8 |
| 12 | LIFE-RAFT-DB-15 | 4-14 person capacity, deep ballast system, canopy, with drogue (average) | 1.91 | 0.78 | 11.9 |
| 13 | LIFE-RAFT-DB-16 | 4-14 person capacity, deep ballast system, canopy, with drogue, light loading | 1.95 | 0.21 | 10.1 |
| 14 | LIFE-RAFT-DB-17 | 4-14 person capacity, deep ballast system, canopy, with drogue, heavy loading | 2.19 | 1.39 | 16.0 |
| 15 | LIFE-RAFT-DB-18 | 15-50 person capacity, deep ballast system, canopy, general (mean values) | 3.68 | 0.34 | 2.8 |
| 16 | LIFE-RAFT-DB-19 | 15-50 person capacity, deep ballast system, canopy, no drogue, light loading | 3.93 | 0.38 | 0.7 |
| 17 | LIFE-RAFT-DB-20 | 15-50 person capacity, deep ballast system, canopy, with drogue, heavy loading | 3.15 | 0.39 | 4.4 |
| 18 | LIFE-RAFT-DB-21 | Deep ballast system, general (mean values), capsized | 0.88 | 0.18 | 11.6 |
| 19 | LIFE-RAFT-DB-22 | Deep ballast system, general (mean values), swamped | 0.99 | 0.14 | 8.0 |
| 20 | LIFE-RAFT-SB-6 | Life-raft, shallow ballast (SB) system AND canopy, general (mean values) | 2.68 | 1.10 | 22.3 |
| 21 | LIFE-RAFT-SB-7 | Life-raft, shallow ballast system, canopy, no drogue | 2.96 | 1.21 | 22.2 |
| 22 | LIFE-RAFT-SB-8 | Life-raft, shallow ballast system AND canopy, with drogue | 2.31 | 0.95 | 22.4 |
| 23 | LIFE-RAFT-SB-9 | Life-raft, shallow ballast system AND canopy, capsized | 1.68 | 0.24 | 8.1 |
| 24 | LIFE-RAFT-SB-10 | Life Raft - Shallow ballast, canopy, Navy Sub Escape (SEIE) 1-man raft, NO drogue | 3.30 | 0.50 | 22.4 |
| 25 | LIFE-RAFT-SB-11 | Life Raft - Shallow ballast, canopy, Navy Sub Escape (SEIE) 1-man raft, with drogue | 1.70 | 0.00 | 23.5 |
| 26 | LIFE-RAFT-NB-1 | Life-raft, no ballast (NB) system, general (mean values) | 3.70 | 1.98 | 28.2 |
| 27 | LIFE-RAFT-NB-2 | Life-raft, no ballast system, no canopy, no drogue | 5.34 | 2.26 | 20.5 |
| 28 | LIFE-RAFT-NB-3 | Life-raft, no ballast system, no canopy, with drogue | 3.15 | 1.51 | 29.2 |
| 29 | LIFE-RAFT-NB-4 | Life-raft, no ballast system, with canopy, no drogue | 3.39 | 1.49 | 23.7 |
| 30 | LIFE-RAFT-NB-5 | Life-raft, no ballast system, with canopy, with drogue | 1.21 | 0.92 | 37.2 |
| 31 | USCG-RESCUE | Survival Craft - USCG Sea Rescue Kit - 3 ballasted life rafts and 300 meter of line | 2.48 | 0.32 | 7.4 |
| 32 | AVIATION-1 | Life-raft, 4-6 person capacity, no ballast, with canopy, no drogue | 3.39 | 1.49 | 23.7 |
| 33 | AVIATION-2 | Evacuation slide with life-raft, 46 person capacity | 2.71 | 0.72 | 14.9 |
| 34 | LIFE-CAPSULE | Survival Craft - SOLAS Hard Shell Life Capsule, 22 man | 3.52 | 1.44 | 22.2 |
| 35 | OVATEK-CRAFT-1 | Survival Craft - Ovatek Hard Shell Life Raft, 4 and 7-man, lightly loaded, no drogue (average) | 3.51 | 1.08 | 13.9 |
| 36 | OVATEK-CRAFT-2 | Survival Craft - Ovatek Hard Shell Life Raft, 4 man, lightly loaded, no drogue | 3.86 | 0.32 | 6.7 |
| 37 | OVATEK-CRAFT-3 | Survival Craft - Ovatek Hard Shell Life Raft, 7 man, lightly loaded, no drogue | 3.26 | 2.32 | 22.3 |
| 38 | OVATEK-CRAFT-4 | Survival Craft - Ovatek Hard Shell Life Raft, 4 and 7-man, fully loaded, drogued (average) | 1.04 | 0.03 | 8.1 |
| 39 | OVATEK-CRAFT-5 | Survival Craft - Ovatek Hard Shell Life Raft, 4 man, fully loaded, drogued | 1.09 | 0.03 | 5.3 |
| 40 | OVATEK-CRAFT-6 | Survival Craft - Ovatek Hard Shell Life Raft, 7 man, fully loaded, drogued | 0.99 | 0.02 | 6.9 |
| 41 | PERSON-POWERED-VESSEL-1 | Sea Kayak with person on aft deck | 1.16 | 0.41 | 10.2 |
| 42 | PERSON-POWERED-VESSEL-2 | Surf board with person | 1.93 | 0.51 | 14.8 |
| 43 | PERSON-POWERED-VESSEL-3 | Windsurfer with mast and sail in water | 2.25 | 0.69 | 11.5 |
| 44 | SKIFF-1 | Skiff - modified-v, cathedral-hull, runabout outboard powerboat | 3.15 | 1.29 | 22.3 |
| 45 | SKIFF-2 | Skiff, V-hull | 2.87 | 0.32 | 0.5 |
| 46 | SKIFF-3 | Skiffs, swamped and capsized | 1.65 | 0.39 | 13.3 |
| 47 | SKIFF-4 | Skiff - v-hull bow to stern (aluminum, Norway) | 2.52 | 1.07 | 10.5 |
| 48 | SPORT-BOAT | Sport boat, no canvas (*1), modified V-hull | 6.54 | 2.19 | 18.5 |
| 49 | SPORT-FISHER | Sport fisher, center console (*2), open cockpit | 5.55 | 2.27 | 22.2 |
| 50 | FISHING-VESSEL-1 | Fishing vessel, general (mean values) | 2.47 | 2.76 | 48.2 |
| 51 | FISHING-VESSEL-2 | Fishing vessel, Hawaiian Sampan (*3) | 2.67 | 2.98 | 48.1 |
| 52 | FISHING-VESSEL-3 | Fishing vessel, Japanese side-stern trawler | 2.80 | 3.13 | 48.2 |
| 53 | FISHING-VESSEL-4 | Fishing vessel, Japanese Longliner (*3) | 2.47 | 2.76 | 48.2 |
| 54 | FISHING-VESSEL-5 | Fishing vessel, Korean fishing vessel (*4) | 1.80 | 2.01 | 48.2 |
| 55 | FISHING-VESSEL-6 | Fishing vessel, Gill-netter with rear reel (*3) | 3.72 | 1.41 | 23.9 |
| 56 | COASTAL-FREIGHTER | Coastal freighter. (*5) | 1.87 | 2.09 | 48.2 |
| 57 | SAILBOAT-1 | Sailboat Mono-hull (Average) | 4.50 | 4.95 | 47.7 |
| 58 | SAILBOAT-2 | Sailboat Mono-hull (Dismasted, Average) | 3.94 | 3.98 | 45.3 |
| 59 | SAILBOAT-3 | Sailboat Mono-hull (Dismasted - rudder amidships) | 6.27 | 4.01 | 32.6 |
| 60 | SAILBOAT-4 | Sailboat Mono-hull (Dismasted - rudder missing) | 2.42 | 0.79 | 18.1 |
| 61 | SAILBOAT-5 | Sailboat Mono-hull (Bare-masted, Average) | 5.23 | 5.89 | 48.4 |
| 62 | SAILBOAT-6 | Sailboat Mono-hull (Bare-masted, rudder amidships) | 6.66 | 7.36 | 47.9 |
| 63 | SAILBOAT-6b | Sailboat Mono-hull (Bare-masted, rudder hove-to) | 2.69 | 2.81 | 46.2 |
| 64 | SAILBOAT-7 | Sailboat Mono-hull, fin keel, shallow draft (was SAILBOAT-2) | 2.67 | 2.98 | 48.1 |
| 65 | SAILBOAT-8 | Sunfish sailing dingy - Bare-masted, rudder missing | 1.79 | 0.10 | 12.2 |
| 66 | FV-DEBRIS | Fishing vessel debris | 1.97 | 0.36 | 10.4 |
| 67 | SLDMB | Self-locating datum marker buoy - no windage | 0.00 | 0.00 | 0.0 |
| 68 | SEPIRB | Navy Submarine EPIRB (SEPIRB) | 0.40 | 0.00 | 70.2 |
| 69 | BAIT-BOX-1 | Bait/wharf box, holds a cubic metre of ice, mean values (*6) | 0.72 | 1.86 | 30.8 |
| 70 | BAIT-BOX-2 | Bait/wharf box, holds a cubic metre of ice, lightly loaded | 2.53 | 1.09 | 13.3 |
| 71 | BAIT-BOX-3 | Bait/wharf box, holds a cubic metre of ice, full loaded | 1.15 | 1.48 | 36.7 |
| 72 | OIL-DRUM | 55-gallon (220 l) Oil Drum | 0.75 | 0.48 | 37.1 |
| 73 | CONTAINER-1 | Scaled down (1:3) 40-ft Container (70% submerged) | 1.78 | 0.27 | 0.8 |
| 74 | CONTAINER-2 | 20-ft Container (80% submerged) | 1.25 | 0.19 | 10.5 |
| 75 | MINE | WWII L-MK2 mine | 1.07 | 0.41 | 19.1 |
| 76 | REFUGEE-RAFT-1 | Immigration vessel, Cuban refugee-raft, no sail (*7) | 1.56 | 0.08 | 8.3 |
| 77 | REFUGEE-RAFT-2 | Immigration vessel, Cuban refugee-raft, with sail (*7) | 6.43 | 2.22 | 20.0 |
| 78 | SEWAGE | Sewage floatables, tampon applicator | 1.79 | 0.16 | 5.1 |
| 79 | MED-WASTE-1 | Medical waste (mean values) | 2.75 | 0.50 | 10.3 |
| 80 | MED-WASTE-2 | Medical waste, vials | 3.64 | 0.67 | 10.4 |
| 81 | MED-WASTE-3 | Medical waste, vials, large | 4.34 | 0.74 | 9.7 |
| 82 | MED-WASTE-4 | Medical waste, vials, small | 2.95 | 0.54 | 10.4 |
| 83 | MED-WASTE-5 | Medical waste, syringes | 1.79 | 0.16 | 5.1 |
| 84 | MED-WASTE-6 | Medical waste, syringes, large | 1.79 | 0.16 | 5.1 |
| 85 | MED-WASTE-7 | Medical waste, syringes, small | 1.79 | 0.16 | 5.1 |