01 — Verdict
The Google Solar API returns no data anywhere in the Dutch Caribbean.
Curaçao, Aruba, Bonaire and Sint Maarten sit entirely outside every Solar API coverage tier. This is not a gap at the edges of a covered region — it is total absence. A production integration built on the Solar API would return an empty result for every address Dynaf serves.
02 — What the Solar API is
A Google Maps Platform product that estimates rooftop solar potential from aerial and satellite imagery. Where it has coverage it is genuinely strong: it returns roof segment geometry, panel counts, annual sunshine hours, solar flux maps and inputs for financial modelling — the work this proof of concept does by hand. It exposes three endpoints.1
- buildingInsights
- Returns insights about the location, dimensions and solar potential of a building.
- dataLayers
- Returns URLs for raw solar information datasets for an area surrounding a location.
- geoTiff
- Fetches rasters with encoded solar information — digital surface model, imagery, annual and monthly flux maps, and hourly shade.
Coverage is published as four downloadable GeoJSON files, one per data tier: HIGH (aerial imagery at 0.1 m/pixel), MEDIUM (high-altitude aerial at 0.25 m/pixel), BASE (satellite imagery at 0.25 m/pixel) and Detected Arrays.2
03 — Coverage point test
Rather than trust a country list, the four official coverage GeoJSON files were downloaded and tested directly with a ray-casting point-in-polygon algorithm. Results below are the tiers that contain each point.
| Location | Coordinates | Tiers returned |
|---|---|---|
| Dynaf markets | ||
| Willemstad, Curaçao | 12.1091, −68.9316 | NONE |
| Sint Michiel, Curaçao | 12.1500, −68.9833 | NONE |
| Westpunt, Curaçao | 12.3667, −69.1500 | NONE |
| Oranjestad, Aruba | 12.5240, −70.0270 | NONE |
| Kralendijk, Bonaire | 12.1500, −68.2833 | NONE |
| Philipsburg, Sint Maarten | 18.0237, −63.0458 | NONE |
| Controls — expected covered | ||
| Amsterdam, Netherlands | 52.3676, 4.9041 | HIGH, MEDIUM, DA |
| San Juan, Puerto Rico | 18.4655, −66.1057 | HIGH, BASE |
| Miami, United States | 25.7617, −80.1918 | HIGH, MEDIUM, DA |
| Control — expected empty | ||
| Mid-Atlantic ocean | 20.0000, −40.0000 | NONE |
Grid sweep — every island, end to end
Single points can miss a small covered patch, so each island's full bounding box was swept at roughly 0.005° (~550 m) spacing. Not one point in any tier.
| Area | Points tested | HIGH | MEDIUM | BASE | Arrays |
|---|---|---|---|---|---|
| Curaçao | 8,585 | 0 | 0 | 0 | 0 |
| Aruba | 2,601 | 0 | 0 | 0 | 0 |
| Bonaire | 4,087 | 0 | 0 | 0 | 0 |
| Sint Maarten | 1,271 | 0 | 0 | 0 | 0 |
| Control — Puerto Rico | 45,981 | 7,750 | 30,186 | 42,231 | 0 |
| Control — Amsterdam | 2,013 | 1,782 | 2,013 | 0 | 2,013 |
How far is the nearest coverage?
| From | Nearest covered polygon vertex |
|---|---|
| Willemstad, Curaçao | 210.4 km (BASE) · 664 km (MEDIUM) · 693 km (HIGH) |
| Oranjestad, Aruba | 90.6 km (BASE) · 663 km (MEDIUM) · 693 km (HIGH) |
| Kralendijk, Bonaire | 280.1 km (BASE) · 644 km (MEDIUM) · 668 km (HIGH) |
| Philipsburg, Sint Maarten | 124.2 km (MEDIUM) · 223.9 km (BASE) · 296 km (HIGH) |
Method. The four official coverage GeoJSON files (dataset stamp 20260506, the current version published on the coverage page as of 2026-08-07) were downloaded and parsed per RFC 7946: the first ring of each polygon is treated as the exterior and subsequent rings as holes, so a point counts as covered only if it falls inside an exterior ring and outside every hole. Testing is pure ray-casting with a bounding-box pre-filter. Positive controls (Amsterdam, Miami, San Juan) and a negative control (open ocean) all behaved as expected, which is what makes the island zeroes trustworthy rather than a parsing bug. File checksums were recorded and re-verified.2
04 — Pricing
Pay-as-you-go per SKU, billed per 1,000 requests, with a free monthly allowance per SKU. Listed for completeness — none of it is reachable from Dynaf's territory today.3
| SKU | Tier | Free / month | 0–100k | 100k–500k | 500k–1M | 1M–5M | 5M+ |
|---|---|---|---|---|---|---|---|
| Building Insights | Essentials | 10,000 | $10.00 | $5.00 | $4.50 | $4.00 | $3.50 |
| Data Layers | Enterprise | 1,000 | $75.00 | $37.50 | $33.75 | $30.00 | $26.25 |
USD per 1,000 requests. Building Insights SKU 1856-4940-856A, Data Layers SKU 54EE-B608-7DDE.
05 — Where the Solar API does fit
Worth knowing, because Dynaf's clients and partners operate outside the islands too.
- Instant residential quoting
- Address in, roof geometry and panel layout out, with no site visit — the mass-market installer funnel.
- Lead qualification at scale
- Score a whole address list for solar suitability before spending sales time.
- Portfolio and municipal surveys
- Estimate generation potential across many buildings from Data Layers flux rasters.
- Shading-aware design
- Hourly shade and monthly flux support realistic layouts on complex roofs.
- Existing-array detection
- The Detected Arrays tier identifies roofs that already have panels — useful for market sizing.
Coverage concentrates in the US, much of Europe, Japan, Australia and parts of Latin America. In the Caribbean it is patchy: Puerto Rico is well covered, most independent islands are not.
06 — Recommendation
- Do not build on the Solar API for the Dutch Caribbean. Not as a primary source and not as a fallback. It returns nothing for every address in Dynaf's territory, and no published timeline says that will change.
- Use the desk-quote method in the other tab as the working path. Operator-traced roof outlines on satellite imagery, combined with PVGIS irradiation, produce a defensible estimate today with no Google dependency and no per-request cost. Every assumption is visible and editable, which matters more for a governance audience than an opaque API result would.
- Monitor coverage rather than wait on it. Google publishes no early-access programme, coverage-request form or regional roadmap — the FAQ says only that "New data are added all the time, but we don't have a specific refresh rate for specific regions."5 The practical move is to re-run the point-in-polygon test against the published GeoJSON on a schedule and get told the day Curaçao appears. That check is the script already written for this assessment; it takes seconds to run.
Stated honestly: raising the gap with a Google Maps Platform sales or partner contact is a reasonable thing to try, but nothing published indicates it influences the coverage roadmap. It is not represented here as a route that is known to work.
07 — Sources
- Solar API overview — endpoint definitions and derived data. Page last updated 2026-07-31 UTC. Accessed 2026-08-07.
https://developers.google.com/maps/documentation/solar/overview - Solar API coverage — quality-tier definitions and the four downloadable coverage GeoJSON files, dataset stamp 20260506. Page last updated 2026-07-31 UTC. Accessed 2026-08-07.
https://developers.google.com/maps/documentation/solar/coverage - Google Maps Platform pricing — Solar API SKU identifiers, volume tiers and free monthly caps. Accessed 2026-08-07.
https://developers.google.com/maps/billing-and-pricing/pricing - Solar API usage and billing — pay-as-you-go SKU model. Accessed 2026-08-07.
https://developers.google.com/maps/documentation/solar/usage-and-billing - Solar API FAQ — coverage refresh statement. Page last updated 2026-07-31 UTC. Accessed 2026-08-07.
https://developers.google.com/maps/documentation/solar/faq - PVGIS v5.3, European Commission Joint Research Centre — irradiation source used by the quote tool. Accessed 2026-08-07.
https://re.jrc.ec.europa.eu/pvg_tools/en/ - NASA POWER — fallback irradiation source. Accessed 2026-08-07.
https://power.larc.nasa.gov/