What this model captures — and what it doesn't
An illustrative simulator, not a forecast. Ranges come from peer-reviewed literature — see citations below. Read these caveats before citing the outputs.
Backwards validation
Rewinding the sliders to their 1973 values (canopy 25%, built-up 15%, water 55 km², vehicles index 3) against the April 2026 baseline should reproduce the IMD-recorded April mean.
Historical temperature · 1951–2024
Source: Open-Meteo ERA5 Archive, IMD Safdarjung. 2 m air temperature reanalysis — not LST. Different from the homepage +LST stat.
Now captured
A monthly offset table (IMD climatology 1991–2020) shifts the baseline temperature to reflect the city's dry-hot, monsoon, and post-monsoon phases. The "Climate context" month selector lets you explore the full annual cycle.
Four cardinal wind directions apply a multiplier to the slider-driven temperature delta based on the land-use axis they blow across — dense built-up / IT corridors warm, green / coastal fringes cool. Multipliers and PM2.5 offsets are calibrated per city where published wind-roses exist and inherited with caveat elsewhere.
An AOD slider (0.1–1.0) captures aerosol radiative forcing in both directions: daytime cooling from solar dimming, nighttime warming from IR trapping. AOD also feeds into PM2.5. A time-of-day toggle switches which effect dominates.
Each city is divided into five zones, each with its own canopy/built-up/water baseline and a residual zone temperature offset. Selecting a zone snaps the sliders to that zone's land-use baseline and adds the zone offset to the output.
Live PM2.5 from CPCB/state-board stations via OpenAQ, updated every 15 min (falls back silently if unreachable). The live reading is observational only — it does not rebase the model's PM2.5 calculation, which remains a function of the slider state against the April 2026 baseline.
Still not captured
Coefficients are calibrated to zone-mean LST from Landsat 30 m studies. Individual streets can differ by 3–5°C depending on tree shade, building geometry, surface albedo, and local traffic.
During winter months, haze plumes from the Indo-Gangetic plain can advect into other regions. The AOD slider captures local aerosol load but cannot simulate multi-day transport events or associated PM2.5 spikes of 200+ µg/m³.
The April 2026 baseline already embeds decades of warming. The sliders explore the urban-heat-island contribution but do not project future years under SSP scenarios.
The live weather strip shows Open-Meteo near-surface air temperature at the city centroid. It is not a land surface temperature (LST) measurement, not disaggregated by zone, and not real-time satellite imagery.
Urban heat island intensity is modulated by boundary-layer height, synoptic cloud cover, and antecedent soil moisture. These require a mesoscale numerical weather model and cannot be reduced to a slider coefficient.
Air conditioning units, data centres, and industrial heat discharge represent a meaningful UHI contribution but have no peer-reviewed city-wide emission inventory at the needed resolution. The vehicle slider proxies road transport only.
Coefficient table
Central estimates from regression studies. Low/high give the reported uncertainty range — propagated as low–high bands next to every central number in the simulator readouts. Currently calibrated to Bangalore; per-city overrides (monsoon, wind, AOD, built-up) land with the validated citations for each city.
| Driver | Change | Central | Range | Effect on |
|---|---|---|---|---|
| Tree Canopy | Per −1 pp canopy | +0.09°C | 0.06–0.12°C | LST (daytime) |
| Built-up Area | Per +1 pp built-up | +0.075°C | 0.05–0.1°C | LST |
| Water Bodies | Per −1 km² water | +0.55°C | 0.3–0.8°C | LST (within 500 m) |
| Vehicles Index | Per +10 pp index | +4 µg/m³ | 3–5 µg/m³ | PM2.5 only |
| AOD forcing (day) | Per +0.3 AOD above 0.4 | -0.8°C | single-value estimate | LST cooling |
| AOD forcing (night) | Per +0.3 AOD above 0.4 | +0.5°C | single-value estimate | LST warming |
Citations
- Ziter et al. 2019 — Scale-dependent interactions between tree canopy cover and impervious surfaces reduce daytime urban heat during summer. PNAS 116(15): 7575–7580. Canopy coefficient.
- Manoli et al. 2024 — Seasonal and diurnal modulation of the urban heat island by tree cover. Nature Communications. Canopy range.
- IISc Ramachandra & Bharath 2023 — Spatiotemporal dynamics of urbanisation and LST in Bangalore 1973–2023. Built-up coefficient and historical land-use data.
- Sustainable Cities & Society 2024 — Meta-analysis of urban water body cooling effects across Asian megacities. Water body coefficient 0.3–0.8°C/km².
- KSPCB / UrbanEmissions APnA 2018 — Bangalore vehicle-fleet emission factors; wind-rose 2022.
- Babu et al., ARFI 2013 — Aerosol Radiative Forcing over India. AOD forcing coefficients.
- IMD climatology 1991–2020 — Monthly mean temperature normals per city. Monsoon and seasonal offset source.