Nighttime Development Analysis for India
Is this area growing or fading? See it in 9 years of NOAA satellite night lights — the same data economists use to track development.
What do night lights tell you about a piece of land?
Every night, NOAA's VIIRS Day/Night Band satellite records how bright every 463-metre patch of Earth is at midnight. That brightness — measured in nanowatts per square centimetre per steradian — is one of the few publicly available signals that tracks where India is actually developing, year on year, with no surveys, no government filings, and no asking permission.
What night-light radiance actually measures
Radiance is the physical brightness of light leaving the ground at night. It is dominated by four things: street lighting and grid electrification, buildings staying lit after sundown (offices, factories, retail, hotels), vehicle headlights on roads, and active industrial operations (kilns, smelters, gas flares). It does not measure population directly. It measures the visible footprint of electrified human activity. HITEC City in Hyderabad went from 41 to 70 nW/cm²/sr between 2016 and 2025 — that 70% jump tracks the IT parks, malls, and 24x7 office towers that came online in that window.
Why growth matters more than absolute brightness
The brightest plots in India are already developed and already expensive. Connaught Place, Bandra-Kurla, MG Road Bengaluru — these have been bright for two decades. There is no upside left in their radiance because there is no upside left in their land prices. The plots where money is being made are the ones moving from dark to bright.
The classic appreciation curve looks like this:
- Year 0: Sparse village, radiance under 1 nW/cm²/sr. Land sells at agriculture prices.
- Year 3–5: Highway access added, basic electrification. Radiance climbs to 5–15. Speculators arrive. Land doubles.
- Year 5–8: First gated communities and warehouses come up. Radiance hits 20–40. Land triples again.
- Year 8–12: Fully developed peri-urban zone. Radiance 40–80. Land prices plateau at urban rates.
The buying window is between year 3 and year 5 — when growth is unmistakable on satellite imagery but local prices haven’t fully caught up. Mahindra City corridor in Chennai, the Outer Ring Road extension in Hyderabad, Sarjapur-Attibele in Bengaluru — all crossed from dark to bright in this exact pattern over 2016–2023.
India case studies
Rapid Urbanisation (>100% growth): Sangareddy and parts of Patancheru on Hyderabad’s western edge crossed from under 5 to over 25 nW/cm²/sr between 2016 and 2025. This pattern — under 5 then crossing 20 — is a reliable signal of an active development corridor.
Strong Growth (50–100%): The Outer Ring Road corridors of every Tier-1 city. The Yamuna Expressway townships near Greater Noida. The new IT zones north of Pune (Hinjewadi Phase III). Each shows the textbook curve from peri-urban to fully developed.
Declining / Stagnant (≤ 0%): Several historical textile towns in interior Tamil Nadu (Karur, Erode) and small mill towns in Maharashtra and Gujarat have flat or falling radiance — local industry has shrunk faster than residential growth has compensated. These are not value buys; they are structural declines.
What it does not tell you (the caveats)
Night-light data is one signal in a stack of signals, not a complete answer.
- Point sources can mislead. A single stadium, large factory, gas flare, port crane yard, or fishing fleet anchored offshore can spike a small parcel reading. Cross-check with the Development Activity tool, which counts actual buildings.
- Tree cover blocks light. Heavily forested high-income gated communities can register lower radiance than they should. Outer Bengaluru’s lake-side enclaves are an example.
- 463m resolution is coarse. A 1-acre parcel falls inside one or two VIIRS pixels. The reading reflects neighbourhood activity, not the specific plot.
- 2016 baselines of zero need care. Newly electrified areas show “infinite” percentage growth — we cap this at 100% and flag the confidence field.
How to spot the pre-peak appreciation window
Three signals stacked together are the strongest buying signal you can get from public satellite data:
- Current radiance between 5 and 25 nW/cm²/sr — not dark, not yet bright.
- 9-year growth above 50% — the trajectory is unmistakable.
- Development Activity tool showing active construction momentum — buildings are going up but the area isn’t saturated.
Plots that match all three are typically 18–36 months ahead of the appreciation wave. Plots that match only one are noise.
What do these numbers mean?
The result card shows where the area is today, where it was 9 years ago, and where the trajectory points. Here's how to read each number.
How the analysis works
Three steps. Under one second. No setup.
You mark the plot
Draw the exact boundary on the map, drop a pin at the centre, or paste an address. For VIIRS readings, plots above 1 km² give the most stable reading — each pixel is roughly 463 m × 463 m.
We sample three annual composites
Your boundary is overlaid on three NOAA VIIRS Day/Night Band annual mean composites — 2016, 2021, and 2025. Each composite averages 365 nights of cloud-masked satellite imagery into one radiance value per pixel.
You see the full 9-year curve
The result card shows your three radiance values — 2016, 2021, and 2025 — the percentage change, and an urbanisation trajectory label derived from the growth rate.
Methodology and data sources
We use public NOAA satellite data, processed locally. No commercial API, no third-party processing service.
Where the data comes from
The single source is the VIIRS Day/Night Band Annual Composite V2.1/V2.2, published by NOAA’s Earth Observation Group at the Colorado School of Mines. We download the cloud-masked, outlier-removed annual mean (average_masked) variant directly — no API key, no rate limit. Each year is one global GeoTIFF at roughly 463 m resolution that we clip to India and store as a tiled COG.
How we score development
For every analysis, we compute two things:
- Current radiance — mean of VIIRS pixel values from the 2025 composite that intersect the input geometry. Because the 50 m point buffer is smaller than one 463 m VIIRS pixel, we use
all_touched=Truein the rasterio mask so any pixel touched by the geometry is included. - 9-year growth —
(radiance_2025 − radiance_2016) / radiance_2016 × 100. Newly electrified areas (2016 baseline = 0) are capped at 100% growth and flagged in the confidence field.
Coverage and limitations
- Spatial resolution is ~463 m per pixel. Plots smaller than 3 pixels (~1.4 km²) get a confidence note — the reading reflects neighbourhood-level activity, not the specific parcel.
- Point sources like stadiums, gas flares, large factories, and ports can spike a small parcel reading. Cross-check with the Development Activity tool for ground truth.
- Tree cover attenuates upward-emitted light. Forested low-density gated enclaves can read lower than their actual development level.
- Annual averaging smooths out one-off events but also smooths out very recent construction. A building lit up in December 2024 will only show in the 2025 composite at roughly 1/12 its actual brightness.
What we don’t do
We do not predict future radiance. The growth percentage is observed historical change, not a forward projection. We also do not directly convert radiance to GDP, population, or property prices — those correlations exist in academic literature but vary by region; we leave that interpretation to the user.