The Sky Doesn't Go Dark in Monsoon: Measuring Light Pollution Through India's Cloudy Season

India's monsoon erases the stars for months at a time — yet the artificial light flooding up into those clouds tells a story that clear skies cannot

It is the third week of July, and you are standing on the terrace of a friend's flat in Bengaluru. The Southwest Monsoon arrived six weeks ago and has not relented. The sky overhead is a uniform, brownish-orange ceiling — not a single star visible, not even Jupiter, which should be blazingly obvious this month. But the ceiling itself is glowing. Entire sections of it pulse with the reflected amber of street lamps and the blue-white wash of a shopping-complex floodlight two kilometres away. The clouds are not dark. They are lit from below, and they are bright.

A student next to you holds up her phone and asks whether SkyQI can still take a reading. You tell her, honestly, that you are not sure. She takes the photo anyway, uploads it, and the result comes back: Bortle 9. SQM below 17.0 mag/arcsec². The platform flags the reading with a cloud-cover warning, but it doesn't discard it. It keeps it.

That result is worth thinking about. It is not measuring the sky. It is measuring what the city is doing to the clouds.

This post is about what cloud cover reveals, what it hides, and how the four monsoon months — June through September across most of India — can still produce meaningful light-pollution data for citizen scientists willing to understand what they are actually measuring.

Why the Monsoon Doesn't Stop Light Pollution

The widespread assumption is that cloud cover and light pollution are separate concerns: one is weather, the other is infrastructure. In fact, clouds and artificial light interact in a way that makes the monsoon season one of the most dramatic light-pollution demonstrations available to anyone who cares to look.

Artificial light travels upward. In a city, a substantial fraction of every streetlamp, billboard, mall facade, and office window directs light sideways and skyward rather than downward onto the surface it is supposed to illuminate. Under a clear sky, much of this light escapes into space after scattering off air molecules — which is what creates the familiar urban skyglow visible from outside cities. Under a cloud layer, something different happens.

Clouds — particularly the low-level stratus and nimbostratus that dominate the Indian monsoon — act as reflectors. Light travelling upward from a city hits the cloud base and bounces back down, where it illuminates the city again, which sends more light back up, creating a feedback loop. The technical term for this is retro-reflection or cloud amplification, and it is why a heavily overcast night in Mumbai or Kolkata is often brighter at street level than a clear night, not darker.

Studies using ground-based photometry have shown that low cloud cover can increase artificial skyglow by a factor of roughly two to ten times compared to clear-sky conditions at the same urban site, depending on cloud height, cloud optical depth, and the distribution of light sources below. In raw SQM terms, a site that reads 19.0 mag/arcsec² on a clear night might drop to 17.5 or even 17.0 under a thick overcast — a shift of 1.5 magnitudes, corresponding to roughly four times more sky brightness.

This is not a new finding. It has been documented in European and North American cities for decades. But in India, it plays out across an entire subcontinent for four months every year, over some of the most densely illuminated urban areas on Earth.

What a Sky Quality Meter Actually Measures Through Clouds

A Sky Quality Meter — and by extension, SkyQI's camera-based analysis — measures the luminance of the sky in units of apparent magnitude per square arcsecond (mag/arcsec²). A higher number means a darker sky. The device does not know why the sky has the brightness it measures. It reports what the sensor sees.

Under a clear, moonless sky at a genuinely dark site, the sky's natural background luminance comes from several sources: airglow (chemiluminescence in the upper atmosphere), zodiacal light (sunlight scattered off interplanetary dust), starlight integrated across the entire sky, and diffuse Milky Way glow. Together these produce a natural sky background of roughly 22.0 mag/arcsec² at the zenith — Bortle 1 territory.

Under a cloud layer with no artificial light at all — say, a monsoon overcast over the Thar Desert far from any settlement — the cloud base itself has some luminance from these same natural sources. A thick cloud layer over a truly dark site will read somewhere around 21.0–21.5 mag/arcsec², measurably brighter than the clear-sky baseline but still in Bortle 2–3 range.

Under a cloud layer over a city, the calculation changes entirely. The cloud base is now illuminated from below by artificial sources. The SQM or camera sees a sky that is glowing with redirected city light. The reading falls — sometimes dramatically. In the dense core of Delhi or Mumbai on a heavily overcast monsoon night, readings below 17.0 mag/arcsec² are common, placing the sky firmly in Bortle 9 territory. Not because the stars are behind clouds, but because the city has turned the clouds into a luminous ceiling.

This is what makes the cloud-amplification effect so significant for light-pollution science. A clear-sky reading tells you how much light escapes upward into space. A cloud-cover reading tells you how much is being redirected back down to the ground.

Note what happens at the dark site versus the bright site as clouds thicken. At the dark site, clouds reduce the SQM reading slightly — the sky gets a little brighter as natural light is scattered back. At the bright urban site, clouds dramatically reduce the SQM reading — the sky gets far brighter as artificial light is amplified. The divergence between the two tells you, quantitatively, how much of a city's upward light output is being returned to Earth by the cloud layer.

The Monsoon Amplification Across Indian Cities

India's monsoon creates an almost uniquely useful natural experiment. The same cloud layer — low, thick, and continuous — settles over cities of wildly different sizes and light-pollution levels simultaneously. By comparing cloud-cover SQM readings across those cities, it becomes possible to rank their upward light-output in a way that is actually more revealing than clear-sky comparisons.

This is a counterintuitive point worth dwelling on. On a clear night, comparing skies from Delhi and Bengaluru is complicated by the fact that Delhi's drier, dustier atmosphere scatters light differently from Bengaluru's wetter air. The measurements are hard to compare directly because the optical medium is different. But under a uniform cloud layer of similar height and thickness, both cities are illuminating the same type of reflective surface. The cloud becomes a calibrated screen, and the brightness of that screen reflects the city's total upward light output fairly directly.

Field observations and satellite studies have consistently shown that the order of Indian cities by upward artificial light output does not always match intuitive assumptions based on population:

• Delhi is large and sprawling, but its light is spread over a very large footprint.
• Bengaluru is, in some measurements, comparably bright or brighter than Delhi in its core zones, despite a smaller population. The city's commercial corridors — Outer Ring Road, the IT clusters of Whitefield and Electronic City — produce extremely high-density lighting.
• Mumbai is bright across a narrow linear footprint shaped by the peninsula, with exceptional density in Bandra, Andheri, and the business districts.
• Kolkata produces intense skyglow over a compact core, with a relatively sharp falloff outside the city.
• Hyderabad has expanded dramatically, with HITEC City and the Gachibowli corridor contributing to a skyglow footprint larger than its pre-2000 extent would suggest.
• Chennai has among the lower upward-light readings of the major metros, though the T. Nagar and Anna Nagar commercial zones remain intensely lit.

These patterns, visible from satellite, become directly measurable from the ground during the monsoon, because the cloud layer creates the reflective surface that makes them legible.

Can SkyQI Measurements Through Clouds Be Trusted?

Honestly: yes and no, and the distinction matters.

A SkyQI reading taken under heavy cloud cover accurately measures the apparent sky brightness at that moment. The number is real and reproducible. What it cannot tell you, without additional context, is how much of that brightness is cloud amplification of artificial light versus intrinsic cloud luminance from natural sources.

In practice, at any site within 50 km of a significant urban area in India, the answer during a monsoon overcast is: almost all of it is artificial. Natural sky background luminance under thick cloud is not bright enough to register below 21.0 mag/arcsec². Anything brighter than that — and certainly anything in the Bortle 6–9 range — under thick cloud cover is city light coming back down.

At a genuinely remote dark site — say, the plains around Hanle, or the interior of the Thar far from Jaisalmer — a thick overcast might read in the 20.5–21.5 range. If a SkyQI reading from such a location under monsoon cloud returns 19.0 mag/arcsec², the algorithm will flag the result as possibly influenced by a distant light source that the observer hadn't noticed, or by a nearby source not identified during upload.

SkyQI handles cloud-affected readings in three ways:

1. Flagging: If the image analysis detects a sky that is uniformly bright without visible stars, and the brightness exceeds a threshold consistent with cloud amplification (roughly below 19.5 mag/arcsec² at any site not already rated Bortle 6 or worse under clear skies), the reading is tagged with a cloud-interference warning. The data is retained but labelled.

2. Seasonal weighting: For long-term site characterisation, the platform uses a simple seasonal filter. Clear-sky readings taken between October and May carry higher weight in computing a site's "baseline" Bortle class. Cloud-cover readings from June through September are stored in a separate partition, analysed for what they reveal about upward light output rather than sky transparency.

3. Delta analysis: The most scientifically interesting comparison is between a site's clear-sky reading and its thick-overcast reading on a night when conditions are otherwise stable. The difference — expressed in magnitudes — is a proxy for the cloud-amplification factor, and therefore for how much of the city's light is directed upward. SkyQI encourages contributors to take paired readings over the monsoon months specifically to build this dataset.

What Monsoon Readings Reveal That Clear Skies Cannot

There is a piece of information that clear-sky photometry simply cannot provide: how much of a city's emitted light is directed upward versus downward.

On a clear night, the light that goes up into space is essentially lost to ground-level measurement unless you have a sky-facing photometer. You can measure the sky's brightness, but the fraction of total municipal light output that accounts for it is not directly observable from the ground.

Under a reflecting cloud layer, however, the picture changes. The cloud sends a fraction of the upward light back to the surface. If you know the cloud's reflectivity (albedo) — approximately 0.6 to 0.8 for thick stratus, based on standard cloud-physics values — you can work backwards from a ground-level measurement to estimate the total upward flux. This is simplified physics, but it is the same reasoning that satellite teams use when they interpret cloud-reflected light in night-time imagery.

What this means for Indian cities is that the monsoon season, far from being a dead period for light-pollution science, is actually the best time of year to estimate upward light emission. A systematic dataset of cloud-cover SQM readings from the same sites, taken over multiple monsoon seasons, would let researchers track whether a city's upward light output is growing or shrinking — independently of any changes in atmospheric clarity.

Changes in street-lighting technology — the large-scale transitions from sodium-vapour lamps to LEDs that have been happening in Delhi, Bengaluru, and other cities over the past decade — should, if managed correctly, reduce total upward emission. LED luminaires, when properly shielded, direct more light downward and less sideways or upward. Whether Indian LED retrofits are actually achieving this is a question that cloud-cover photometry can help answer, because LEDs and sodium lamps have different spectral signatures that show up differently in the colour temperature of cloud-reflected glow.

The orange-amber ceiling of pre-LED cities — familiar to anyone who grew up near an Indian highway — is being replaced by a harder, bluer-white glow in areas where LED streetlamps have been installed. That colour shift is visible in SkyQI images, and it is worth documenting.

Practical Monsoon Measurement: A SkyQI Guide

If you are reading this in June, July, August, or September and wondering whether to bother contributing measurements, the answer is: yes, emphatically, but with a few adjustments to how you think about what you are measuring.

Do not aim for dark-sky photography. You will not see stars. You are not supposed to. The goal shifts entirely to measuring the luminance of the cloud base.

Choose a zenith-pointing shot, not a horizon shot. The cloud layer is thickest and most uniform at the zenith. Horizon shots during the monsoon pick up city glow on the near horizon, which confounds the measurement.

Note the cloud type if you can. High cirrus behaves very differently from low stratus. Cirrus is thin and poorly reflective; stratus and nimbostratus are thick and highly reflective. A reading under cirrus will not show the same amplification as one under dense monsoon overcast. The SkyQI upload form has a cloud-cover field — use it, and note "thick overcast", "thin cloud", or "partial cloud" as appropriate.

Pair each overcast reading with the most recent clear reading from the same location. If your Bengaluru terrace measured 18.2 mag/arcsec² on a clear May night, and now reads 16.8 under thick July cloud, that delta — 1.4 magnitudes — is quantitatively meaningful. It represents roughly a 3.6× increase in sky luminance from cloud amplification, which tells you something about the fraction of your neighbourhood's lighting that is directed upward.

Take readings at the same time of night. Light pollution is not constant over a night's duration. Retail districts are brighter at 9 PM than at 2 AM; industrial areas may be the reverse. For paired clear-vs-cloudy comparisons to be meaningful, the time of night should be consistent — midnight is a reasonable standard because it captures both residential and commercial sources while avoiding peak-evening transient spikes.

Repeat across the monsoon season. A single reading has limited value. A reading per week over sixteen weeks creates a time series that reveals whether particular events — a festival like Ganesh Chaturthi, Navratri, Diwali on its fringes, or a major sporting event — produce detectable spikes in the cloud-reflected glow.

What This Means for SkyQI Readings

SkyQI's value as a citizen-science platform rests on the density and continuity of its data. Gaps in the dataset — like the four months when most contributors stop submitting because they assume overcast skies are useless — are not just inconvenient. They make it structurally harder to answer questions about whether Indian cities are getting darker or brighter over time, because a year's worth of data has a systematic seasonal hole in it.

The monsoon months are also, ironically, the months when some of the most socially significant light-pollution events occur. Ganesh Chaturthi pandals in Mumbai and Pune are lit with extraordinary intensity. The run-up to Navratri and Dussehra brings temporary commercial lighting at scales that dwarf normal retail illumination. And across North India, the pre-Diwali commercial season begins in earnest in October — just as the monsoon retreats and the skies start clearing.

Cloud-cover readings taken during the festival buildup period, even while the tail of the monsoon lingers, capture these spikes in a way that clear-sky readings taken a week later, when the decorations are down, cannot. That documentation matters.

There is also a quieter use for monsoon data. SkyQI's long-term goal is to build a seasonal light-pollution baseline for every district in India — a map that shows not just how dark the sky is on a single good night, but how it varies across the year, through weather, through festivals, through the slow accumulation of new infrastructure. That map cannot exist without twelve months of data per year. The monsoon season is four of those twelve months. It is not optional data.

Tonight: Measuring the City's Own Light

If it is raining outside, or if the sky is a solid, featureless orange-grey ceiling, do this: go to the darkest patch of your roof, terrace, balcony, or garden that you can reach. Point your phone at the zenith. Take the photo. Upload it to SkyQI.

Note what time it is. Note whether the cloud cover is thin, moderate, or thick. Note whether you can see the cloud base clearly or whether it blends into the general glow.

Whatever number comes back, it is not telling you that you live under a Bortle 9 sky — it is telling you what your city's upward-directed light looks like when the atmosphere hands it back to you. The sky is not a blank during the monsoon. It is a mirror.

Every reading you contribute this season adds a data point to a seasonal record that no telescope, no satellite, and no government monitoring station is currently building for Indian cities at this resolution. The Vedic astronomers who compiled the Surya Siddhanta knew that the sky had patterns invisible on any single night but legible across a year's careful watching. The monsoon clouds are not an obstacle to that watching. They are, if you know how to read them, another instrument.

Measure tonight. The map needs the monsoon too.