Are India's big cities getting hotter?

Are India's big cities getting hotter?

Ask whether India's big cities are getting hotter and the honest answer comes in layers. Yes, the country has clearly warmed, and almost every city has warmed with it. But the part you would expect to rise fastest, the blistering afternoon, often isn't the part that's rising. And the dry thermometer, the one number everyone quotes, turns out to hide the most dangerous trend of all: how hot the air actually feels once you account for humidity.

This page builds the answer from 38 cities, not four, and three lenses of heat. First the national backdrop from two independent records. Then annual averages, very hot days, and hot nights, city by city. Then the lens the dry thermometer misses entirely: feels-like heat. That last one is where the story turns.

Start with the whole country, not the cities

Before judging any single city, it helps to know whether the background is warming at all. It is. Berkeley Earth's homogenised record, stitched from thousands of station histories back to the nineteenth century, shows India's average temperature rising by more than a degree. Each stripe is one year, blue cool, red warm, and the run of deep red on the right is the last two decades. In 2016, the warmest year in the record, India sat about 1.06°C above its 1951–1980 average.

That is the climate-change backdrop. Cities sit on top of it, and many add their own heat through concrete, traffic, and lost greenery. So the real question is not whether warming is real. It is how that warming shows up, street by street.

A second dataset, the same upward line

One record is never enough, so the second chart uses an entirely different method. ERA5, the Copernicus reanalysisReanalysis (ERA5)A method that runs every past observation, satellites, balloons, ships, stations, through a weather model to produce a complete, physically consistent record, even where no thermometer existed. ERA5 is the leading one, from Copernicus and ECMWF.Both the national line and all 38 city series are built this way, which is why coverage reaches back to 1940, but it is a modelled estimate, not one street's thermometer. run by the European Centre for Medium-Range Weather Forecasts, blends satellites, weather balloons, and surface readings into a physically consistent estimate of past weather. Its India-wide annual mean rises from about 23.2°C in the 1940s to 24.2°C in the most recent decade, close to 1°C of warming, arrived at independently of Berkeley Earth.

Two methods, two teams, one upward line. That agreement is what lets the rest of this page zoom into individual cities with confidence. The city series below are built from the same ERA5 reanalysis, read at each city's coordinates.

Do the cities all move together?

This is where four cities was never enough. The grid shows annual average temperature for 38 cities, each as its own small chart from 1940 to 2025. Read them together and a pattern appears: almost every line drifts upward, but the slopes differ and the year-to-year noise is real. Hill-valley Srinagar and the southern cities climb steeply; the dense northern plains wobble more.

A single city's first-year-to-last-year comparison can mislead, because any one year can be a fluke. The fix is to compare averages, not endpoints, which is what the next chart does.

Which cities warmed most, and the two that didn't

This chart shows each city's warming directly, its 2016–2025 average minus its 1940s average, ten years at each end so a single flukey season cannot drive the result. Bars run right for warming, left for cooling, sorted largest first.

Almost all of them point right: 36 of the 38 cities are warmer than in the 1940s. Kochi tops the warming at about +2.2°C, with Srinagar, Thiruvananthapuram, Coimbatore, and Madurai all near or above +1.8°C, the south and the mountains warming fastest.

The exceptions are revealing. Only two bars point the other way, and they are neighbours: Delhi (about −0.4°C) and Amritsar (about −0.7°C), both on the irrigated northwestern plains. That is not a data error, and not a reprieve from summer heat. It is the documented "warming hole" of northwest India, where decades of irrigation and heavy haze have damped the daytime average even as the climate around it warms. Hold that thought, these two cities come back later, because the dry thermometer is lying about them.

Very hot afternoons send no single signal

Now switch from averages to the afternoons people dread: days when the maximum reaches 35°C or more. You might expect these to rise everywhere. They don't.

By the same decade-average measure, very hot days climb sharply in some cities, Madurai gains about 44 a year, Kochi 42, Guwahati 26, Coimbatore 25, but fall in others, including Dehradun (about −30), Chennai (−15), Ahmedabad (−14), and Amritsar (−14). Read the grid and you see no single direction.

The reason is the threshold, not the climate. A fixed 35°C line flatters cities that already sail past it and barely moves where afternoons were always extreme; on a humid coast a "cooler" 33°C afternoon can be more dangerous than a dry 38°C one, yet it never trips the count. The afternoon-peak metric is real, but it is the wrong place to look for the clearest warming signal.

Hot nights are the signal that travels

The nights are different. This chart counts nights when the temperature never drops below 28°C, when the body gets no overnight break to recover. Unlike very hot days, hot nights are rising in almost every city.

Chennai's count roughly doubled, from about 43 nights a year in the 1940s to 93 in the recent decade, touching 113 in 2024. But this is not only a coastal story. Surat gained about 42 nights, Patna 36, Ahmedabad 32, Madurai and Vijayawada about 31 each, with cities across the north and east all moving up. Even Delhi's nights crept up while its daytime average did not.

Warm nights are the heat metric most tightly linked to illness and death, because sleep without cooling is what steadily wears the body down. That this signal rises so broadly, coastal and inland, north and south, is one of the two findings this page is built around.

The cities where the night stopped cooling

The year-by-year lines for the worst-hit cities show how recent the change is. Chennai's climb steepens sharply after the 2000s; the inland and coastal cities bend upward in the same window. This is a late-twentieth-century acceleration, not a slow nineteenth-century drift.

So far, though, every chart has used the dry thermometer, the temperature of the air. For how heat actually feels to a human body, that is only half the story. The other half is humidity.

The thermometer isn't the whole story

Feels-like temperature, what meteorologists call apparent temperature, folds humidity, wind, and sunshine into the dry reading. Humid air stops sweat from evaporating, so the body can't shed heat; 34°C in steamy Kochi punishes you far more than 34°C in dry Jodhpur. This grid shows annual mean feels-like temperature for all 38 cities, from the same ERA5 reanalysis.

Here the story turns. Feels-like temperature is rising faster than the dry thermometer almost everywhere. Kochi's feels-like mean is up about 3.2°C since the 1940s, against 2.2°C dry. Madurai and Thiruvananthapuram are up close to 2.9°C, Coimbatore about 2.8°C. And remember Delhi and Amritsar, the two cities whose dry averages didn't rise? Their feels-like temperatures still climbed, about +1.9°C in Delhi and +1.6°C in Amritsar. The dry thermometer said they got a pass. Once you account for the moisture in the air, they didn't. It was hiding the danger.

How dangerous the air actually feels

To make that concrete, this chart counts "dangerous-heat days", days when the feels-like maximum reaches 40°C or more, the level at which heat starts to threaten healthy people. Cities are ranked by how many such days they now average, with the change since the 1940s alongside.

The rise is steep and wide. Madurai gained about 96 dangerous-heat days a year; Kochi 74; Guwahati 70; Kolkata 63, coast, deep south, and Northeast alike. Delhi, supposedly the city that "cooled," went from about 82 such days to 112. The cities at the top of the ranking now spend a third to nearly half the year in feels-like-40°C territory. This is the wet-bulb dimension earlier versions of this page couldn't show, and it is the most alarming chart here.

Where humid heat is climbing fastest

Finally, the year-by-year lines for the fastest-rising cities show how recent and how steep the humid-heat climb has been, Madurai, Kochi, and Guwahati bending sharply upward in the satellite era.

This is the takeaway to leave with. India's cities are getting hotter, but "hotter" is mostly two quieter stories than the afternoon headline: the night that no longer cools, and the air that increasingly feels worse than the thermometer admits. Both are humidity-shaped, both are health stories, and both are rising across almost every city we measured, including the two whose dry temperatures pretended nothing had changed.

Sources and how we measured this

This page draws on three open datasets. The opening warming stripes come from Berkeley Earth (a homogenised station record, shown as anomalies versus the 1951–1980 average, through 2020). The national average line and all 38 city series come from ERA5, the Copernicus / ECMWF reanalysis, read at each city's coordinates through Open-Meteo, annual values built from daily data, 1940–2025, complete calendar years only (a partial 2026 is excluded so a half-year can't masquerade as a cool spell).

Every "change" figure compares a city's 1940s average against its 2016–2025 average, ten years at each end, so that no single hot or mild year can manufacture a trend. Thresholds are fixed: a very hot day is a daily maximum of 35°C or above; a hot nightHot night (tropical night)A night when the temperature never falls below 28°C. It matters more for health than a hot day because the body relies on cooler nights to recover; without that break, heat stress builds up.Hot nights rise across almost every city, making them, with humid heat, the article's central finding. is a minimum of 28°C or above; a dangerous-heat dayDangerous-heat dayHere, a day when the feels-like maximum reaches at least 40°C, the level at which heat begins to threaten even healthy people. It is a humidity-aware cousin of the dry 'very hot day' (35°C maximum).It is the closest this page gets to a wet-bulb stress measure, and it rises far more consistently than the dry afternoon count. is a "feels-like" maximum of 40°C or above. Feels-like (apparent temperature) folds humidity, wind and sunshine into the air temperature.

A few honest limits. A reanalysis is a modelled estimate, not a single neighbourhood's thermometer, and each city is read as one grid point, so local heat islands, informal settlements and workplace exposure aren't captured. Feels-like is modelled, not a clinical wet-bulb reading. In the small-multiple grids, each city's chart is auto-scaled to its own range, so read the slope, is it rising?, rather than the height; the coloured number beside each city is the actual change. And Berkeley Earth's record ends in 2020 while the ERA5 series run to 2025, so end dates differ by source.