Seasonal ocean forecasts can now flag elevated marine heatwave risk months ahead — sometimes. Here's what skill exists, where it fails, and who uses it.
If you manage a kelp restoration site, a salmon farm or an oyster lease, the question is brutally practical: will next summer be another 2014–2016, another hot blob, another wiped-out season? Two decades ago the honest answer was "we'll know when the buoys scream." Today the answer is more interesting: seasonal prediction systems genuinely can flag elevated marine heatwave risk months in advance — in some regions, in some seasons, with skill that beat climatology. The art is knowing where the forecast is sharp and where it is still fog.
This explainer lays out what actually gets predicted, why some coasts are easier than others, what "skill" really means in this field, and how a restoration or farming operation can use probabilistic outlooks without being misled by them. For background on the phenomenon itself, start with Marine Heatwaves 101.
A marine heatwave (MHW) is usually defined relative to the local seasonal norm: a period of five or more days when sea-surface temperature exceeds the 90th percentile of the climatology for that time of year. Forecasting one is therefore not like forecasting rain on Thursday. Operational outlooks typically deliver probabilities — for example, "a 60% chance of MHW conditions in this region over the coming three months" — derived from ensembles of coupled ocean–atmosphere models run out weeks to months ahead.
There are three time horizons, each with distinct skill:
The single biggest lever on MHW predictability is how much of the local temperature variability is driven by a predictable climate mode.
The tropical Pacific is the star pupil. El Niño and La Niña — the beating heart of seasonal forecasting (see El Niño and kelp) — reorganise the atmosphere and ocean on timescales of months to a year, and modern systems predict El Niño state with useful skill two to three seasons ahead. Regions whose heatwaves ride on El Niño teleconnections — the tropical Pacific itself, and to a meaningful extent parts of the eastern Pacific boundary — inherit some of that skill.
Western boundary currents are moderately predictable. Regions like the Tasman Sea extension of the East Australian Current, the Kuroshio, and the Agulhas are driven by large-scale wind patterns with their own slow rhythms; research has shown seasonal MHW skill in the Tasman region of months, because the ocean there "remembers" its state via ocean heat content rather than just surface temperature.
Eastern boundary upwelling coasts are hard. Off California, Humboldt's Chile or Namibia, surface temperature is the tug-of-war outcome between winds, upwelling and regional currents on week-to-week timescales. Skill is typically modest beyond a season, and the local forecasts that matter most are short-range: will the wind relax next week and let warm surface water pool against the kelp?
Persistence is the quiet workhorse. Warm anomalies, once established, decay slowly — months for deep mixed layers. A large share of operational "skill" anywhere is simply carrying today's anomaly forward, which is why the hottest tool in the box is often a good map of what the ocean is doing right now.
Under the hood, a seasonal MHW outlook is an ocean–atmosphere model ensemble:
Advances now arriving include machine-learning forecast systems trained on reanalysis data, which have shown competitive skill at weeks-to-months horizons, and rapid MHW "attribution and alert" services that classify ongoing events by likely driver.
A responsible reading of the current science:
For councils, farmers and restoration teams, the forecast is an input to risk management, not a verdict. Practical habits:
Read probabilities as odds, not promises. A 60% MHW chance is a "load the dice" signal, not a date. Actions with low regret — pre-positioning monitoring gear, scheduling the dive season's critical measurements early — become clearly justified; actions with high regret still need a second look.
Match the action to the horizon. Days-to-weeks skill supports operational moves (delaying a hatchery intake, moving a survey window). Seasonal skill supports strategic ones: holding seed stock in reserve, timing seeded-line deployment to ride out an early-season warm spell, or fast-tracking harvest before a forecast warm anomaly stalls recovery markets.
Prefer ensembles and trend over any single run. Many public outlooks come with accessible probability maps — pair them with live buoy data (our guide to reading ocean buoy data shows how) so you can watch a forecast verify or fail in real time and react on the short-range horizon.
Plan your refugia in advance. Forecasting matters most when you know where your coast's cool pockets are. Thermal refugia — upwelling-fed coves, deep-rafted canopy edges, fjord sills — are the places a bad season hurts least; mapping them ahead of time converts a forecast into a triage plan.
Three developments are worth watching. First, higher-resolution regional ensembles will push useful skill down to the scales of individual bays and islands — the scales where restoration actually happens. Second, ecological forecasts: translating a temperature probability into a biological one ("chance of canopy loss at this site," "urchin recruitment risk"), marrying ocean models with ecosystem-response data. Third, decadal planning layers — not predictions, but robust risk maps for where heatwave frequency will cross ecological thresholds over the coming decade, informing where infrastructure and protected areas should go.
The kernel of hope: the ocean is far more predictable than the atmosphere that torments it, because it moves slowly and remembers. Every Argo float, every moored sensor line, every verified hindcast improves the odds. Prediction will never make a marine heatwave harmless — but the difference between a warning and a surprise, for a forest that can lose half its canopy in a season, is nearly everything.
Useful probability forecasts exist at 1–3 months for large parts of the ocean, with the best skill in the tropics (linked to El Niño/La Niña) and some boundary-current regions. Day-to-week forecasts of local temperature, currents and upwelling are considerably sharper. Precise timing of a local extreme months ahead remains beyond any system.
An El Niño forecast predicts a basin-scale Pacific climate state; an MHW forecast predicts the probability of local temperature extremes, which may or may not be driven by El Niño. El Niño is one driver — a big, predictable one — but many damaging heatwaves (like the Northeast Pacific "Blob") had weak or complex ties to it.
Extremes usually require several drivers to stack — background warming, a climate mode, a blocking weather pattern, weak winds thinning the surface mixed layer. Ensembles capture elevated probability of each ingredient but tend to under-sample the worst combinations, so forecasts get the direction right more often than the magnitude.
Yes. Several national meteorological and ocean services publish seasonal MHW outlook maps and regional ocean forecasts openly, and short-range ocean forecasts are embedded in many marine weather apps. The key skill is reading probabilities: use them to break ties in decisions and schedule low-regret actions, not as deterministic schedules.
Indirectly but meaningfully. Seasonal outlooks inform when to deploy seed stock, when to pre-position monitoring, and when to expect urchin or disease pressure to shift. At Tidewright's fictional sites we treat a flagged warm season as a cue to prioritise surveys at known thermal refugia and delay vulnerable life-stage work — the same logic any operator can apply with public outlooks.
Almost certainly. Better subsurface ocean observing (more Argo-class floats), higher-resolution regional models, and machine-learning systems trained on decades of reanalysis are all trending in the same direction. The harder limits — chaos in the atmosphere and the sheer compounding of extremes — will keep seasonal forecasts probabilistic rather than precise for the foreseeable future.
Marine heatwaves are weeks of anomalously hot ocean that can deforest entire coastlines. Here's the science behind them and what restoration can do.
Dr. Lena Fischer ·
Every few years the Pacific reorganises itself and kelp forests pay the price. How El Niño starves kelp — and how La Niña's cool afterwash lets it recover.
Dr. Mara Ellison ·
Over 90 percent of the planet's trapped heat is in the ocean. How we measure ocean heat content, from expendable probes to 4,000 drifting floats — and why it matters.
Dr. Lena Fischer ·