2026 has a 93% shot at being the hottest year ever recorded

Heat announces itself long before anyone puts a number on it. Pavement softens, overnight lows stop offering relief, hospital admissions climb among people who spent a shift in conditions their bodies could not shed, and utilities watch demand push past what their grids can carry. A hot summer registers first as a lived thing, in water restrictions and canceled practices and the hum of air conditioners that never cut off. What it does not come with is a ranking. Nobody standing in a heat wave knows whether the week belongs among the hottest ever measured or sits inside the ordinary range, and the difference matters well beyond trivia. Insurers reprice coverage. Power companies size generation capacity for the summers ahead. Public health agencies decide where to put cooling centers and how early to open them. The answer arrives only after the measurements come in.

Confirming a global heat milestone takes an accounting that no single thermometer can supply. Weather stations crowd some regions and leave others thin, ships and buoys sample an ocean covering most of the planet's surface, and the historical comparisons that give a sweltering month its meaning reach back to an era when instruments and practices differed from today's. A credible answer has to reconcile all of it, then attach an honest margin of error to the result. The reconciliation carries weight far outside science, since governments negotiating emissions targets, water managers setting allocations, and farmers choosing what to plant all take their cues from where the trend line sits. A second complication landed this year in the Pacific. The ocean's largest natural warming cycle began gathering strength partway through the calendar, feeding additional warmth into a climate system already running hot. The cycle's arrival turns a straightforward tally into a forecasting problem. Does 2026 mark a peak, or the opening of something larger that carries into the next year?

Berkeley Earth publishes a monthly global temperature analysis that answers questions of exactly that kind. For its August 2026 report, the group combined two large datasets: a high-resolution land temperature record built from more than 22 million monthly station readings collected at upward of 53,000 locations, and an ocean surface record adapted from the UK Met Office Hadley Centre's HadSST4 archive, which draws on over 500 million shipboard and buoy measurements. Blending the two produces a single global grid at quarter-degree resolution, and each resulting figure comes with a margin of error reflecting gaps in older records, instrument limitations, and changes in measurement practice over the past 170-plus years. Every comparison in the report measures against average conditions from 1850 to 1900, the period climate scientists treat as a proxy for pre-industrial temperatures and the reference point written into international climate targets. The report covers August 2026 on land and at sea, the northern summer as a season, the state of the Pacific cycle alongside what forecast models expect from it, and the odds that 2026 finishes first in the annual accounting. Figures for the most recent month stay provisional, since observations continue to arrive after publication, though such revisions tend to be small enough to leave the broad conclusions intact.

Illustration: Quartz

August 2026 closed at 1.71 degrees Celsius above the 1850-1900 average, the warmest August in a global series that opens in 1850. Only September 2023 stands higher across every month in that archive, Augusts and otherwise. August 2023 held the previous high for the month and sits about 0.10 degrees Celsius below. The month became the second ever to clear 1.7 degrees above pre-industrial conditions and one of just 10 to pass 1.6 degrees. August marked the second month in a row above 1.5 degrees, after a cool opening to the year in which no month reached the top of the all-time list.

Land surfaces contributed less to the outcome than their usual dominance would suggest. Continental temperatures averaged 2.34 degrees Celsius above the same baseline, second among Augusts and so close to August 2024 that the two amount to a tie. Land has behaved this way throughout 2026, hugging the long-term trend line and declining to push into new territory the way the oceans have.

Ocean surfaces carried the month. Water temperatures averaged 1.296 degrees Celsius above the 1850-1900 benchmark, topping every other August at sea and every other calendar month as well. Across 2,120 months of marine measurement, none has run hotter, and August 2026 slips past August 2023 for first place by a slim margin. Sea surface temperatures have climbed as the equatorial Pacific heated across much of its width without a break.

The planet's actual temperature, seasonal cycle included, averaged 17.16 degrees Celsius in August, a shade under July 2023 at 17.21 and above July 2026 at 17.14. Earth's thermometer crests in July most years, a rhythm that the concentration of land north of the equator dictates. An August that overtakes its own July shows how much additional energy the developing El Niño pumped into the system over the course of the summer. Oceans set their all-time monthly high before that Pacific event peaked, which places the ceiling for sea surface temperatures somewhere beyond August.

Illustration: Quartz

El Niño drove the Niño 3.4 index to an August average near +2.6 degrees Celsius, with daily readings in the central Pacific surpassing anything logged since at least 1940 and reaching those highs months before the usual seasonal crest. The climb has not been smooth. Values flattened for much of August and finished under what forecast models had projected, a pause that hinted the event might be running out of fuel. Westerly wind bursts ended the lull, and the index resumed its ascent in early September.

The August figure holds near 2 degrees Celsius even after the arithmetic accounts for the more than 1 degree Celsius rise in the ocean's mean temperature, the level that marks the entry point to the strongest tier of events. A warmer baseline inflates raw anomalies that reference older ocean conditions, and correcting for it sets current Pacific warmth alongside past episodes of similar force instead of in a class of its own. Even on the adjusted scale, nothing in the archive matches what the central Pacific is doing this early in the year, though conditions stop short of an all-time high.

Forecasters have converged on a record-strength outcome. Every dynamical model in the September ensemble projects one by the standard measure, the Oceanic Niño Index, and the odds of a record on the relative version, which discounts the long-term heating of tropical waters, sit near 98%. Agreement at that level is rare in seasonal forecasting, where models split over timing and magnitude.

The median of 14 seasonal forecasting models puts the crest a fraction above +4 degrees Celsius in the central Pacific between November 2026 and January 2027, more than a full degree beyond the previous high-water mark. Events of that caliber have rearranged atmospheric circulation and redistributed rainfall across whole continents, drenching some regions and leaving others parched. The global temperature response trails the Pacific peak by several months, carrying the bulk of this event's heating past the end of the calendar year. The Pacific has run up a bill that comes due in 2027.

Illustration: Quartz

The August assessment gives 2026 a 93% chance of closing the year as the hottest on record, a verdict drawn from the first eight months of the calendar. In June, the same arithmetic put those odds at 12%. July's revision raised them to 69%. A Pacific event gathering strength and oceans running warmer than at any comparable point in the record explain the leap. A runner-up finish remains the only alternative carrying real weight, at roughly 7%, and everything below it adds up to less than 1%.

The likely annual figure for 2026 falls between 1.536 and 1.642 degrees Celsius above the 1850-1900 baseline, a span that places a final result above 1.5 degrees beyond serious doubt. Through August the year has averaged 1.489 degrees, trailing only one other January-to-August period in the entire dataset, with the chilly opening months costing it the top spot. The most recent 12-month window, September 2025 through August 2026, averages 1.482 degrees, and only the comparable stretches closing in 2024 and 2025 run hotter, a figure that had sat near the planet's decades-long warming path before the strengthening El Niño pushed it upward once more. No month before August claimed an outright high of its own, most finishing second, third, or fourth in their categories, so the annual outcome rests on heat sustained across the calendar instead of one towering spike.

Oceans and continents part company in the annual odds. A record yearly ocean average has become a near certainty, while land holds odds of about 2.5% of reaching its own annual high. The likeliest land result, at roughly 87%, leaves 2024 holding that title. Land started 2026 well below its eventual warmth, and the El Niño boost arriving now falls short of closing that gap before December arrives. The gap separating the two surfaces has grown wider since last month's reading, and four months of data remain before the books close. So the year can claim the hottest global average ever measured while the continents where people farm, build, and work never touch a peak of their own, a divide that locates 2026's defining heat offshore.