GUNIF

Geospatial Urban Nexus Intelligence Framework Raipur, Chhattisgarh · Sharma et al.

IndicesHeat PM₂.₅LULC

Overview

Correlation

LULC

PM_2.5

Recommendations

Data Sources

Layers

Scenarios

3D Twin

Study area

City	Raipur, Chhattisgarh, India
Lat / Lon	21.25 °N, 81.63 °E
Area	~226 km² (municipal)
Indices	NDVI, NDBI, NDWI
Heat	LST, SUHI, Heat Retention
Air quality	25 junctions + 3 CPCB stations

How to use

Toggle layers from the panel on the top right of the map. When you activate a raster, its legend appears at the bottom-right of the map. Click any junction marker for PM_2.5 statistics. Use the ruler to measure distances.

Key headlines

Index & Layer Guide

NDVI	Normalized Difference Vegetation Index — measures green cover density; higher values (→1) indicate dense vegetation, key cooling driver (β = −5.42 °C/unit).
NDBI	Normalized Difference Built-up Index — quantifies impervious surface intensity; strongest positive LST driver in the model (β = +14.18 °C/unit).
NDWI	Normalized Difference Water Index — detects surface water bodies and moist land; weakly negative correlation with LST (β = −3.80, n.s.).
LST	Land Surface Temperature — satellite-derived thermal surface temperature (°C); ranges 27.53–47.83 °C across Raipur (Δ 20.3 °C).
SUHI	Surface Urban Heat Island — LST anomaly relative to rural reference; delineates zones of localized urban overheating beyond background temperature.
Heat Retention	Composite index of nocturnal heat storage capacity; high values in industrial/dense-built zones reflect slow post-sunset cooling.
Building Density	Count of built structures per km² grid cell from GHSL; significant positive predictor of LST (β = +0.0004, p < 0.001).
USI	Urban Stress Index — composite of Urban Infrastructure (UI), Environmental Stress (ESI), and Environmental Benefit (EBI); five tiers from Minimal to Critical.
PM₂.₅	Fine particulate matter (µg/m³) measured at 25 traffic junctions; city mean 67.4 µg/m³ — 1.7× the NAAQS annual limit of 40 µg/m³.

Modelling & validation

LULC change	Built-up +47.33 km² (+31.7%), Green −11.16 km², Barren −35.81 km²
LST gradient	27.53–47.83 °C (Δ 20.3 °C)
USI exposure	~559 k residents (38%) in High–Critical zones; 286 k (19.6%) in Critical alone
OLS R² (LST)	0.498 (full model); GWR local R² up to > 0.90 in industrial NW
LULC accuracy	87.3 % overall, κ = 0.84
LST validation	Spearman ρ = 0.81 vs IMD records

OLS regression — LST predictors

Full model (N = 511, R² = 0.498, adj. R² = 0.493, F = 100.3, p < 0.001)

Predictor	β	SE	t	p
Intercept	38.23	0.350	109.2	***
NDBI	+14.18	1.507	9.41	***
NDVI	−5.42	2.632	−2.06	*
NDWI	−3.80	2.644	−1.44	n.s.
Building Density	+0.0004	7.4×10⁻⁵	5.12	***
PoI count	−0.084	0.409	−0.21	n.s.

* p<0.05 *** p<0.001. VIF: NDBI = 3.5, NDVI = 14.0 (multicollinearity noted).
Reduced model (NDBI + NDVI only): R² = 0.472, NDBI β = +15.86 (p<0.001).

Pearson correlation (cell-wise)

Indices & PoI counts on a regular grid (N≈10k).

Key Correlation Insights

NDVI ↔ NDWI (r = −0.87): Strong negative coupling — vegetated pixels are rarely water-logged; both independently suppress LST.
NDBI ↔ LST (r = +0.68): Strongest positive link in the matrix — built-up intensity is the dominant thermal driver, confirming OLS β = +14.18.
NDVI ↔ LST (r = −0.46): Moderate cooling effect of vegetation; spatial heterogeneity (GWR) amplifies this in the NW industrial corridor.
NDVI ↔ NDBI (r = −0.58): Expected trade-off — urban expansion directly displaces green cover, creating a compounding heat feedback loop.
Building Density ↔ LST (r = +0.22): Significant but weaker than NDBI — structural density adds thermal mass beyond surface imperviousness alone.
PoI ↔ all indices (r ≈ 0): Point-of-interest counts show near-zero correlation with thermal/spectral indices, consistent with OLS n.s. result (β = −0.084).
NDWI ↔ NDBI (r = +0.20): Weak positive association, likely reflecting peri-urban water bodies co-located with industrial zones along the Kharun river.

Land-use / Land-cover change

Class-wise area (km²): 2017 vs 2024.

Mean PM_2.5 by junction (2025)

NO₂ vs PM_2.5

USI Exposure — current snapshot

Urban Stress Index across 512 grid cells (1 km²), 2024.

Stress tier	Cells	Est. pop.	% pop.
Critical	39	~286 k	19.6 %
High	74	~273 k	18.7 %
Moderate	118	~401 k	27.5 %
Low	167	~360 k	24.7 %
Minimal	114	~139 k	9.5 %

Double-burden cells (Critical USI + healthcare desert): 47. See Scenarios tab for BAU projections to 2040.

What Raipur should NOT do

Cautions drawn from the GUNIF/USI analysis (Sharma et al., 2025).

Do not permit further unplanned built-up expansion along the NH-6 / Bhilai north-western corridor — already the highest-PM_2.5, highest-SUHI, highest-USI zone in the city.
Do not approve industrial estate extensions in Urla–Siltara–Mandhar without mandatory ≥ 30 % green-buffer land-bank conditions; ESI already overwhelms EBI in this belt.
Do not convert the remaining barrenland (only 41.44 km² left after a −46% decline) directly to impervious cover — these patches retain residual cooling capacity and ground-water recharge.
Do not rely on the 5 existing CPCB stations alone — the buffer-gap analysis shows the highest-emission junctions (Tatibandh, PachpediNaka, MakeinIndia) sit outside the 2 km monitoring envelope.
Do not plan new schools, hospitals or housing inside the 47 "double-burden" cells (Critical USI + healthcare desert) without parallel green-buffer / air-shed mitigation.
Do not treat vegetation cover as ornamental — NDVI is quantitatively the strongest negative driver of LST (β = −0.31, p < 0.001) and ranks above road density in the GWR cooling model.
Do not assume one-size-fits-all interventions: GWR shows the NDBI→LST coefficient varies 2.3× across the city, weakest in Naya Raipur (planned, with green buffers), strongest in the unplanned industrial NW.

What Raipur SHOULD do

Priority greening zones. Target the 47 double-burden cells in the industrial NW & eastern highway fringe for urban forestry, mandatory perimeter green-belts (≥ 100 m around industrial estates), and arterial roadside tree-planting.
Monitoring network densification. Add ≥ 4 CPCB-grade stations in the Siltara–Urla–Mandhar belt and at least 10 low-cost sensors at the top-PM_2.5 traffic junctions to close the buffer gap.
Healthcare expansion. Prioritise PHC / CHC placement in the double-burden zones under the National Health Mission; locate facilities to minimise Euclidean distance from Critical-USI cells.
Transport-corridor regulation. Within 2 km of NH-6, enforce stricter BS-VI compliance, freight-vehicle time windows, and a non-motorised-transport (NMT) priority corridor along the Kharun riparian.
Naya Raipur as the template. Replicate its green-buffer + low-FSI envelope in any new peri-urban expansion zone — GWR confirms it materially attenuates the NDBI → LST signal.
Cool-roof & permeable-surface mandate. Heat-retention zones overlap industrial-shed clusters — incentivise high-albedo roofing and permeable pavement retrofits.
Institutional GUNIF cell. Establish a standing municipal-level GUNIF/USI cell to update the indicators annually (open data only — ESRI LULC, Landsat 8, OSM, WorldPop, GHSL), feeding into Master-Plan revisions and the State Action Plan on Climate Change.

Where to act first (priority order)

Siltara–Urla industrial corridor — Strong UHI + PM_2.5 hotspot (99% Gi*).
NH-6 / Bhilai axis — directional growth corridor, 2.3× faster built-up gain within 2 km.
Tatibandh / PachpediNaka / MakeinIndia junctions — top PM_2.5 stations, outside monitoring envelope.
Central commercial core (Jaistambh / Pandri / Kalibadi) — high PoI & activity density, moderate heat retention.
Kharun riparian — protect last contiguous green corridor; formalise as a no-development buffer.

Scenario modelling (see Scenarios tab): Under BAU trajectory, LST city mean is projected to rise +0.84 °C by 2030 and +1.52 °C by 2035 relative to 2024 — driven primarily by NDBI intensification (OLS β = +14.18). Mean PM₂.₅ is projected to breach 80 µg/m³ by 2030. Moderate greening intervention can cap the LST increase to +0.38 °C by 2035.

Research transferability: The GUNIF/USI framework is designed for open-data portability — all indicators can be replicated for any Tier-2 city using ESRI LULC 10m, Landsat 8, OSM, WorldPop, and GHSL without proprietary data. See Data Sources tab for full provenance.

Scenario-backed recommendations

Derived from OLS projections (β_NDBI = +14.18, β_NDVI = −5.42) and LULC trajectory modelling. Each recommendation is directly traceable to a quantified future divergence between BAU and intervention scenarios.

Enforce a hard urban-expansion cap of ≤ 3.4 km²/yr (half the BAU rate). This single zoning instrument is the largest lever in the model: it keeps projected LST rise to +0.70 °C by 2035 versus +1.52 °C under BAU — a 54% reduction in thermal penalty — and prevents ~37 km² of additional impervious surface from forming within the decade.
Achieve and hold green cover above 200 km². The OLS model quantifies vegetation cooling at β_NDVI = −5.42°C per NDVI unit. BAU loses ~9 km² of green by 2035; the Aggressive Green scenario recovers ~20 km². Protecting existing canopy is analytically equivalent to removing roughly 1.5 NDBI units of built-up pressure. Immediate action: gazette the 47 double-burden cells' green patches as no-conversion zones under NRDA authority.
Act before 2028 or face a PM₂.₅ lock-in. The BAU trajectory crosses 80 µg/m³ city-wide mean around 2029–2030 — twice the NAAQS annual standard of 40 µg/m³. Once this threshold is crossed, reversing it requires sustained multi-year intervention. Moderate Green keeps the 2035 mean at ~78 µg/m³; Aggressive Green holds it flat at ~67 µg/m³. Install ≥ 4 CPCB-grade stations in the Siltara–Urla belt immediately to close the monitoring gap before the window narrows.
Prioritise the industrial NW corridor above all other zones. GWR shows the NDBI→LST coefficient is 2.3× higher there than the city mean — meaning the same built-up expansion causes more than twice the heat increase. BAU 2035 projects this corridor reaching +3.5 °C above 2024 baseline. A mandatory ≥ 30% green-buffer land-bank around Urla–Siltara–Mandhar estates, if enacted now, reduces this to ~+1.2 °C under the Moderate scenario — avoiding ~820 k residents entering critical thermal stress conditions by 2040.
Redirect barrenland (~41 km² remaining) to managed green, not impervious cover. BAU converts most of this residual buffer to built-up by 2035. Scenario modelling shows that retaining barrenland as transitional green (even unmanaged) preserves groundwater recharge and contributes ~0.3 NDVI units city-wide — equivalent to ~1.6 °C cooling offset in the OLS model. Statutory classification as a "green reserve" under the Raipur Master Plan revision is the required policy mechanism.
Invest in healthcare infrastructure in double-burden zones now — conditions worsen nonlinearly. Under BAU, the ~47 cells combining Critical USI and healthcare-desert status will expand to an estimated ~60–65 cells by 2030 as thermal stress zones spread and population pressure grows at 2.5%/yr. Locating new PHC/CHC facilities in these cells now costs ~40% less (fewer competing land uses) than after 2030, when built-up infill will have saturated available parcels.
Replicate Naya Raipur's green-buffer + low-FSI model in every new peri-urban expansion zone. GWR confirms Naya Raipur materially attenuates the NDBI→LST signal. Under the Aggressive Green scenario — which operationalises this template city-wide — the difference in projected population at High/Critical USI risk is ~250 k residents fewer by 2035 compared to BAU (~609 k vs ~862 k). Each new peripheral township approved without green buffers forecloses this benefit permanently.
Establish a standing GUNIF monitoring cell with annual update triggers. The model's predictive power (OLS R² = 0.498; GWR local R² > 0.90 in the industrial NW) is sufficient for annual early-warning. A city-level GUNIF cell should flag when built-up growth rate exceeds 4 km²/yr, NDVI falls below 0.22 city-wide, or any junction PM₂.₅ exceeds 85 µg/m³ — each representing a scenario branch-point that shifts the trajectory from Moderate toward BAU. All triggers are measurable from open satellite data (ESRI LULC + Landsat 8) at zero marginal cost.

Primary data sources

All datasets used by the GUNIF pipeline (Table 1 of the manuscript). Open / open-licence wherever possible — the framework is portable to any comparable Tier-2 city.

Pre-processing & harmonisation

Projection: WGS 84 / UTM 44N (EPSG:32644).
Spatial resampling to a harmonised 30 m raster grid (1 km analytical grid for population & USI).
Landsat 8 thermal bands → LST via NDVI-based emissivity correction (Guha et al., 2020).
Traffic junctions extracted via OSM road-graph topological node-degree analysis; 500 m + 1 km buffers for spatial co-occurrence with PoIs.
2 km buffer around the study boundary mitigates edge effects.
Composite indices min-max normalised; USI = UI + ESI − EBI with PCA-weighted UI sub-components.

Open licences

ESRI LULC 10m — ArcGIS Living Atlas (open access).
Landsat 8 OLI-TIRS — USGS EarthExplorer (public domain).
OpenStreetMap road & PoI data — Open Database Licence (ODbL).
WorldPop 2020 — Creative Commons Attribution 4.0.
SRTM 30 m DEM — NASA/USGS, public domain.
PM_2.5 — author-deployed low-cost sensor network at each traffic junction (data available on request).

Future Scenario Prospecting

BAU — Business As Usual
Raipur continues its 2017→2024 trajectory: built-up expands +6.76 km²/yr, green cover shrinks −1.59 km²/yr. No new policy interventions. Industrial NW corridor grows unchecked. This is the cost-of-inaction baseline.

Moderate Green — Managed Growth
Urban expansion rate is halved (~3.4 km²/yr) through zoning reform and infill incentives. Green cover is stabilised (no further net loss). Equivalent to implementing the manuscript's Tier-2 priority recommendations.

Aggressive Green — Nature-First
Net-zero peripheral expansion (infill only). Active urban forestry and riparian restoration adds +10% green cover by 2040. Models the upper-bound benefit of a Naya Raipur-style green-buffer mandate city-wide.

Projections are derived from the OLS regression (β_NDBI = +14.18, β_NDVI = −5.42; R² = 0.498) combined with the measured 2017→2024 LULC trend rates.

Horizon year: 2035

202520282031 203420372040

At-a-glance comparison

BAU reference KPIs

Δ LST above 2024

—

Mean PM₂.₅

—

Green cover

—

Pop. at risk MOHFW & Census projections

—

LULC area snapshot at selected year

LST trajectory (°C above 2024)

Mean PM₂.₅ (µg/m³)

Green cover (km²)

Population at High+Critical USI (thousands)

What you see on the map

The map layers represent measured 2023/2024 conditions. Use the table below as a reading guide: under each scenario, these layers would shift as described. Activate a layer from the Layers tab, then refer here for its projected trajectory.

Active Layer	BAU 2035	Aggressive 2035
NDVI	Shrinks ~8.8 km² further; peri-urban loss accelerates	+10% net gain; riparian corridor restored
SUHI	Strong UHI area expands ~15–20% (NW + NH-6 axis)	Moderate–Neutral zones hold; NW corridor partially cooled
LST (°C)	City mean +1.52°C; industrial NW up to +3.5°C (GWR 2.3×)	City mean +0.05°C; near-neutral change
Heat Retention	Very High Retention zone grows with built-up footprint	Stable or declining; cool-roof mandate effect
NDBI	Positive values spread south & east; NW saturates	No perimeter expansion; core densification only
Traffic Junctions PM₂.₅	Tatibandh/PachpediNaka projected ≥100 µg/m³	Stable at 2024 levels with BS-VI + NMT corridor

Methodological note: Projections use OLS model coefficients and linear extrapolation of 2017–2024 LULC trends. PM₂.₅ uses compound growth proportional to built-up expansion (BAU: 3.1%/yr; Moderate: 1.5%/yr; Aggressive: 0%). Population at risk is derived from population data procured and analyzed on the basis of projections from MoHFW and Census of India; base-year values cross-validated against WorldPop 2020 estimates. GWR spatial heterogeneity is not captured — industrial NW hotspot values will exceed city means by 2.3×. These are indicative planning bounds, not deterministic forecasts.

Layer opacity

Toggle layers on/off and drag the slider to fade them for cross-layer comparison (e.g. find junctions inside the Strong-UHI zone).

Layer symbology

Scenario projection guide

Use this quick-reference alongside active map layers. Values shown for the year selected in the Scenarios tab (or 2035 default).

3D Digital Twin

Interactive 3D rendering of building heights across Raipur's urban footprint, powered by deck.gl.

Building Height Statistics (2023)

Sampled buildings	33,229
Height range	0.5 – 53.0 m
Mean height	6.7 m (~2 storeys)
Source	Google Open Buildings + GHSL 2023
Roads	OSM primary / secondary / tertiary

Height Distribution

■ < 3 m	kiosks / boundary walls
■ 3–6 m	dominant 1–2 storey fabric
■ 6–10 m	2–3 storey residential
■ 10–15 m	mid-rise residential
■ 15–30 m	commercial / institutional
■ > 30 m	high-rise landmarks

Press T anywhere to toggle · Drag to orbit · Scroll to zoom · Hover buildings for details.