Climate-decadal activity viability
Screen how a spot's activity viability changes across the next 20-80 years under multiple climate scenarios. An exploratory research direction backed by the global CMIP6 ensemble. Scenarios are plausible pathways, not forecasts, and results are regional-scale context, not site-level truth.
Where this sits
Exploratory research direction, not a shipped product. Multi-decadal climate viability screening is an exploratory direction on a long horizon, not a generally-available product or a contracted report service. No reports have been produced for clients, nothing here is grant-filing-ready, and there is no self-serve delivery. Regional CORDEX downscaling is not built, and local bias-correction is not fitted or validated for any location today, so a projection runs on the global CMIP6 ensemble and is labelled regional context. Every number in these docs is an illustrative shape. For the research framing, see the scenario viability screening overview.
The endpoints below exist as a research surface and are gated behind a contracted tier. They are described here for reference; treat the request and response shapes as a research sketch, not a live report product.
What it does
For a given (spot × activity × scenario × decade), the engine would run an ensemble of CMIP6 model outputs through the same scoring curves as real-time. Returns: lower / mean / upper score bands, frequency of favorable-or-better days, and dimension-level attribution (does the spot degrade because of wind shifts, water temperature, swell direction?). Because regional downscaling and validated bias-correction are not built, these outputs describe regional-scale scenario change, not a deterministic forecast for an individual site or season.
POST /v1/projections
One spot, multiple scenarios, multiple decades.
{
"spot": {
"point": { lat: 36.013, lng: -5.604 },
"activity": "kitesurfing",
"spotId": "tarifa-balneario"
},
"scenarios": ["SSP2-4.5", "SSP5-8.5"],
"decades": ["2030s", "2050s", "2070s"]
}spot.point (not location) holds the { lat, lng } pair. scenarios accepts up to 4 of SSP1-2.6 | SSP2-4.5 | SSP3-7.0 | SSP5-8.5; decades (not horizonDecades) accepts up to 8 of 2020s through 2090s. The response includes a policy.resolvedProfileSlug and a fixed climate-uncertainty disclaimer string on every call.
POST /v1/projections/portfolio
Same shape as single, but with an array of spots (up to 50). Use for "how might our 30-spot resort network screen in 2050 under SSP3-7.0?" as a regional-scale scenario comparison. Heavy compute, typically 30-60s wall time; consider async + polling for big portfolios.
POST /v1/projections/adaptation-report
Combines portfolio projections with a qualitative narrative, identifying which dimensions are likely to be the binding constraint per spot (wind shift, water temperature, snow line moving up, precipitation regime change). This is a research sketch, not a report Goable delivers today: no adaptation reports have been produced for clients, and nothing it returns is grant-filing-ready or decision-grade.
Sub-spot bias correction
Raw CMIP6 is too coarse for spot-level scoring. The engine has a sub-spot bias-correction mechanism (Bayesian-shrunken multiplicative + additive offsets per variable, propagated through the projection), but it is applied only where it has been fitted and validated against local observation history. It is not fitted or validated for any location today, so projections run bias-neutral and say so. On /v1/projections/adaptation-report the response surfaces resolvedSubSpotSlug so you can see which sub-spot resolved; the plain /v1/projections endpoint instead reports policy.resolvedProfileSlug. Where there is no fitted correction, the engine returns the bias-neutral projection plus an explicit wider-uncertainty signal in confidence.