Cook Islands Sun · map studies · back to the thirteen Mauke styles
One style, fifteen islands — every inhabited and uninhabited island in the country. The Watercolour Botanical sheet, style 04 of the Mauke set, rebuilt as an island-agnostic renderer and run over the lot. Same paper, same washes, same italic hand, same masthead. Click any of them to see it full size.
Nothing here is traced by eye. Every shoreline is stitched from OpenStreetMap coastline ways and every relief band is contoured from the Copernicus GLO-30 surface model, both reprojected into UTM — zone 4S for the eastern islands, the same grid the Mauke survey is held in, and zone 3S for Palmerston and the western atolls, which would otherwise carry a 0.6% error in a printed scale bar. Where OpenStreetMap carries a reef, the reef on the sheet is that reef; where it does not, the sheet says so in its own source note.
Three of the fifteen have a topobathymetric LiDAR survey whose rendered output is held in this repository — Rarotonga, Aitutaki and Mangaia, in ici-lidar-map-images/. Those are pictures, not data, so nothing has been traced off them; they have been used to check the drawn sheets, which is what they are good for. The reef triangle on the Aitutaki sheet and the makatea ring on the Mangaia sheet both match their LiDAR renders.
The other twelve have no survey available to this project at all. Their shorelines come from OpenStreetMap, which for the Cook Islands is largely digitised from satellite imagery and from LINZ hydrographic charts. That is good enough to print and not good enough to navigate by.
| Island | Reef on the sheet |
|---|---|
| Rarotonga | mapped — OpenStreetMap reef or lagoon geometry |
| Aitutaki | mapped — OpenStreetMap reef or lagoon geometry |
| Atiu | indicative — 300 m fringing band, no published survey |
| Mangaia | indicative — 320 m fringing band, no published survey |
| Mauke | indicative — 280 m fringing band, no published survey |
| Mitiaro | indicative — 280 m fringing band, no published survey |
| Manuae | mapped — OpenStreetMap reef or lagoon geometry |
| Takutea | mapped — OpenStreetMap reef or lagoon geometry |
| Palmerston | mapped — OpenStreetMap reef or lagoon geometry |
| Penrhyn | mapped — OpenStreetMap reef or lagoon geometry |
| Rakahanga | mapped — OpenStreetMap reef or lagoon geometry |
| Manihiki | mapped — OpenStreetMap reef or lagoon geometry |
| Pukapuka | mapped — OpenStreetMap reef or lagoon geometry |
| Nassau | mapped — OpenStreetMap reef or lagoon geometry |
| Suwarrow | mapped — OpenStreetMap reef or lagoon geometry |
The land area each sheet draws is the sum of its coastline rings. On the high islands that lands on the published figure almost exactly — Mauke 19.2 km² against a surveyed 19.2, Rarotonga 67.7 against 67.4 — which is the cheapest available proof that the ring stitching is right.
On the atolls the first attempt was not close at all. Penrhyn drew 15.3 km² of land against a published 9.84. The cause turned out to be interesting rather than annoying: OpenStreetMap traces every pale shape in the satellite imagery as coastline, and inside a big lagoon most of those are coral heads awash, not motu. Penrhyn had 268 rings and 163 of them were sitting out in open water.
Rather than guess which were which, the pipeline measures them. Copernicus sees a vegetated motu standing three to five metres up and sees a coral head at sea level, so a ring is drawn as land if it stands above 2.5 m, if it is the island's main landmass, or if a mapper has put a name inside it — and as sand if it does not. That single test reproduces the published land areas across the whole group:
| Island | Published | Drawn as land | Difference | Drawn as sand |
|---|---|---|---|---|
| Rarotonga | 67.4 km² | 67.7 km² | +0% | — |
| Aitutaki | 18.1 km² | 18.2 km² | +1% | 0.0 km² · 5 features |
| Atiu | 26.9 km² | 28.5 km² | +6% | — |
| Mangaia | 51.8 km² | 48.4 km² | -6% | — |
| Mauke | 19.2 km² | 19.2 km² | -0% | — |
| Mitiaro | 22.3 km² | 22.9 km² | +3% | — |
| Manuae | 6.2 km² | 7.1 km² | +15% | 0.1 km² · 3 features |
| Takutea | 1.2 km² | 1.2 km² | +1% | — |
| Palmerston | 2.6 km² | 2.8 km² | +6% | 0.1 km² · 13 features |
| Penrhyn | 9.8 km² | 9.9 km² | +1% | 5.3 km² · 222 features |
| Rakahanga | 4.1 km² | 3.9 km² | -6% | 0.0 km² · 2 features |
| Manihiki | 5.4 km² | 5.1 km² | -5% | 0.2 km² · 31 features |
| Pukapuka | 3.0 km² | 3.9 km² | +28% | 0.5 km² · 11 features |
| Nassau | 1.2 km² | 1.2 km² | +3% | — |
| Suwarrow | 1.6 km² | 1.0 km² | -36% | 0.1 km² · 9 features |
Penrhyn now draws 9.9 km² against a published 9.8, and Aitutaki 18.2 against 18.1 — from a test that never saw either published figure. The two rows still flagged are honest disagreements rather than errors: published areas for Pukapuka range from 1.3 km² (the inhabited motu alone) to over 5 km² (all three), and Suwarrow's islets are so small and so low that a 30 m elevation grid is at the edge of what it can resolve. Nothing is discarded either way — the low features are all still drawn, in sand rather than green, because that is what they are.
The Copernicus grid is a surface model. On a low wooded island it reads the top of the canopy as ground; on a steep one its 30 m posting rounds the point off the ridge. Both errors are real and they run in opposite directions, so rather than hand-wave it, every sheet states its own residual against the published summit. The whole table:
| Island | Published | Copernicus | Residual |
|---|---|---|---|
| Rarotonga | 652 m | 503 m | -149 m |
| Aitutaki | 124 m | 121 m | -3 m |
| Atiu | 72 m | 87 m | +15 m |
| Mangaia | 169 m | 188 m | +19 m |
| Mauke | 28 m | 40 m | +12 m |
| Mitiaro | 11 m | 29 m | +18 m |
| Manuae | 5 m | 20 m | +15 m |
| Takutea | 6 m | 25 m | +19 m |
| Palmerston | 6 m | 24 m | +18 m |
| Penrhyn | 5 m | 24 m | +19 m |
| Rakahanga | 5 m | 23 m | +18 m |
| Manihiki | 4 m | 26 m | +22 m |
| Pukapuka | 5 m | 18 m | +13 m |
| Nassau | 9 m | 20 m | +11 m |
| Suwarrow | 4 m | 19 m | +15 m |
The relief bands are contoured at percentiles of each island's own land surface rather than at fixed heights, so a canopy offset shifts every band together and the shape survives intact. Read these sheets for shape. Do not read a height off one.
Every sheet is fitted to A3, so Takutea at 1.2 km² fills the same page as Penrhyn's 233 km² lagoon. That is deliberate — it is the only way the set reads as a set — but it means you cannot compare two sheets by eye. Each carries its own scale ratio and its own bar, and says so on its face. None of them is for navigation.
Everything editorial — the name, the traditional name, the epithet, the three lines of prose, the scale bar, how much sea room the sheet leaves — lives in one file, tools/islands/config.ts. Change it there and re-run bun tools/islands/build.ts. Geometry is never edited by hand; if a shoreline is wrong, the fix belongs in OpenStreetMap and then in bun tools/islands/fetch.ts --refetch.
bun tools/islands/fetch.ts — pull geometry and relief (cached)bun tools/islands/fetch.ts --refetch — ignore the OSM cachebun tools/islands/build.ts — render all fifteen and this pagebun tools/islands/build.ts atiu — just oneTraditional names are set with the ʻokina where one belongs — ʻEnuamanu, Aʻuaʻu Enua, Araʻura — but they come from OpenStreetMap and from general reference, not from a gazetteer. Pukapuka needs a Pukapukan speaker specifically: its language is not Cook Islands Māori. A Cook Islands Māori speaker should check every one of the fifteen before print, on the same principle that sends the nine approximate Mauke pins back to Mauke. Each sheet says so on its face.
Mauke appears here for completeness, drawn from the same open sources as the other eight so the set is internally consistent. It is not the sheet to print: the survey-grade Mauke watercolour, built from the 2022 LiDAR and the 2017 cadastral sheet, is style 04 on the main contents page and is better in every respect.
No landowner or parcel data appears on any of these sheets, on the same reasoning that keeps it off the Mauke set. No sheet invents a settlement, a road or a reef: the uninhabited islands — Manuae and Takutea — carry no villages because they have none.