Ballygunge

Kolkata · Urban Core · Ward 68

Urban Core

92.6ha inside the ward boundary

22.523° N · 88.369° E

Under ConstructionHeat field · modelledBlock detail · illustrative

— buildings

SELECTING ENGINE

  • Live basemap, real footprints & Google heights
Ballygunge —°C surfaceUrban Core
Baruipur —°C surfacePeri-Urban Fringe
Barrackpore —°C surfaceIndustrial River Corridor
Bengaluru →
Temperatureforecast
Air now

—°C air

—°C feels

—% RH · — m/s

Ground & rooftop surface · modelled nowmodelled
—°C
Δ vs all-green ref
—
Surface above 40 °C
—
What the Δ is, and is not

Modelled land-cover thermal contrast: this ward's mean surface temperature minus a fully-vegetated reference cell under identical forcing. NOT a measured urban heat island — published Kolkata surface UHI is 0.85–1.5 °C from 1 km satellite data, a different quantity.

Surface temperature distribution°C
2632374248
ComfortableWarmHotSevereExtreme

Land surface temp · the scale moves with the weather

What the colours meanmodelled
ComfortableWarmHotSevereExtreme

Building colour: continuous ramp, or snapped to the 5 legend classes

Satellite-checked surface temperature, not air temperature.

How this was checked
  • Night±3.0 °C over 50 ward-scenes of NASA ECOSTRESS. In-sample error 2.662 K against a best possible 2.166 K.
  • Day±4.5 °C, indicative only: n = 29, an earlier evidence set.
  • BandCovers the leave-one-overpass-out error (2.777 K at night), not the in-sample fit.
Caveats OBOS adds when they apply
  • DawnIn the Now view from solar 05:00 to 09:30 the chip reads "Outside validation · ~±7.5 °C at this hour": OBOS keeps 7.54 K for that window, the larger of its out-of-sample measurements of the morning ECOSTRESS stratum, because the surface is still releasing stored heat while the sun is already loading it.
  • SeasonWhen it applies, the reading adds: "Dry-season reading: the modelled surface runs at night ~1.7–1.9 °C warm in Mar–Apr against ECOSTRESS".
What is measured, what is modelled
  • Measured · ward boundary, footprints, satellite scenes
    Solid marks
  • Modelled · the heat field on the map
    Outlined marks
Scenariosmodelled

Not built yet. The toolbox under Overview runs one scenario at a time and keeps none of them: there is nowhere to save an intervention set, name it, or put two of them side by side.

The scoring model it would draw on is already here — src/scripts/climate-engine/dc-urs-scenario.ts — which is why this pane says so instead of rendering an empty panel.

Resilience ScoreComing soon

Withdrawn on 17 Sep 2026. No score is shown until it returns.

Rooftop solar · installable, whole ward

Installable capacity

screened
—MWp

Annual yield
—
Worth at tariff
—
Roofs ≥ 3 kWp
—
Losing ≥ 5 %, of those
—
Ward estimateestimate
Weather nowno reading

Air temperature

—°C

Feels like calculated

—°C

Humidity
— % RH
Wind
— m/s
Sky
—
Visibility
—
Rain memory

    Open ring: no rain or drizzle in that report. The heat model reads these reports to decide whether the ground is still wet.

    What the heat model does with it
    • The airport sets "now". Air temperature, humidity, wind and cloud from the newest report drive the modelled surface.
    • Older than 90 minutes, and the Met Norway forecast stands in, carrying any rain the airport reported.
    • Rain at the airport is applied to the ward, and the weather line says conditions over the ward may differ.
    Ground & rooftop surface modelled

    —

    . A surface, not the air.

    Ward recordprototype

    Ballygunge Urban Core · Ward 68

    What is measured, the out-of-sample error and sample size, every source dataset with its licence, and the analysis CRS and bounding box.

    Ward area
    0.93 km²
    Buildings
    2,207 of 7,931 drawn
    Study square
    1.8 km a side
    Analysis CRS
    EPSG:32645
    Download ward record ↓
    How sure
    • Measuredland surface temperature (LST) — the radiometric temperature of the ground and roof surfaces, as an infrared satellite sees it. Not: air temperature, human thermal comfort, UTCI or any comfort index, indoor temperature. Surface temperature can diverge from air temperature by many degrees in daytime sun; at night the two track closely. Comfort additionally depends on humidity, wind and radiant exchange, none of which this model resolves. Do not read these values as what a person feels.
    • Nightcalibrated · ±3.0 °C · n = 50 · model error 2.662 K · ceiling 2.166 K
    • 13:00 peakindicative · ±4.5 °C · n = 29. Earlier evidence set, measured before Ballygunge became KMC Ward 68: Ballygunge was then the 1.4 km box around 22.528 N, 88.366 E. Not re-adopted; the current measurement over Ward 68 is in /uncertainty and model-accuracy.json strata.peak_ecostress.
    • Patternr 0.27 block by block, below a vegetation map's r 0.31 · n = 85. Ward-level surface temperature is calibrated against ECOSTRESS. The pattern WITHIN a ward is not: block by block it scores r = 0.27, still below the r = 0.31 of a vegetation map given the same treatment, and coarsening the comparison does not close that gap at any scale. At neighbourhood scale (~300-500 m) it reaches r = 0.4, but only about 0.2 in KMC Ward 68, whose dense, uniform fabric leaves the satellite little pattern to match. The colour range inside a ward is close to what the satellite measures (0.97x), though about 0.8x in Ward 68, where contrasts are slightly understated. Ward figures are measured, neighbourhood contrast is indicative, block-by-block detail is illustrative.
    • Heights28 of 30 buildings needed · underpowered. Measured on the earlier 1.4 km Ballygunge box, before Ballygunge became KMC Ward 68; its ICESat-2 transects stop short of Ward 68's southern blocks. Not re-measured.
    Building footprints and every licence
    • 5,582 · OpenStreetMap
      traced by people
    • 1,997 · Google Open Buildings
      machine-detected
    • 352 · Microsoft ML Buildings
      machine-detected

    Share-alike: this polygon, and any adaptation of it, may be redistributed only under CC BY-SA 2.5 India (or a later or compatible CC BY-SA licence), with this attribution. It is not covered by the ODbL that governs the rest of this record.