Climate Vulnerability Assessment

LST, Air Temperature and Heat Stress Indicator Estimator for Delhi

This dashboard is a Google Earth Engine web app for Delhi where a user can selects a year (2015–2026) and month of their interest from drop-down, then clicks Run  which triggers the full analysis pipeline automatically. On Run, it pulls Landsat 8/9 for Land Surface Temperature (LST), Sentinel-2 (or Landsat fallback) for NDVI indices, and ERA5-Land hourly reanalysis for 2 meter air temperature, dew-point, wind, and solar radiation. These coarse ERA5 fields (~11 km) are statistically downscaled to 30m using LST, NDVI, NDBI, MNDWI, and SRTM elevation as predictors via Linear Regression, Random Forest, or Gradient Tree Boost  whichever gives lowest MSE (Mean Square Error). From the downscaled fields (Ta30, RH30, Wind30, Solar30), three heat-stress indices are derived: WBGT, UTCI, and Heat Index.
A Heat Vulnerability Index (HVI) is computed using four weighting methods and auto-selected based on maximum correlation with LST. Results appear as 30m raster maps on the map panel, along with ward-level zonal summaries, a pixel inspector (click any point for values), and a year-on-year comparison tool  giving stakeholders a consistent spatial tool to identify which wards face highest heat risk for any selected month.

The data in the tool is updated with the same frequency as Landsat satellite data (usually with a lag of about one to two months).

Abbreviations: Elev- Elevation above mean sea level; LST- Land Surface Temperature; NDVI- Normalized Difference Vegetation Index; Ta30- Air Temperature at 2 meter height; HVI- Heat Vulnerability Index; HI-Heat Index; WBGT-Wet Bulb Globe Temperature; UTCI- Universal Thermal Climate Index; ERA5- ECMWF atmospheric reanalysis version 5

IMD Forecast for Delhi

Tabular ward-wise data for monthly LST and UTCI (2015-2026)

Click here to access

Technical Documentation

(Click on the hyperlink to download the full detailed technical documentation html file with the following table of contents expanded).

Platform: Google Earth Engine (JavaScript API)
Study Area: Delhi, India — all administrative wards
Analysis Period: 2015–2026 (monthly resolution)
Document Language: English
Document Purpose: Complete technical reference for researchers, scientists, urban planners, Policy makers and science journalists

Table of Contents

  1. Dashboard Overview — What, Why, and Where

  2. Glossary of All Terms — Definitions, Interpretation, Ranges, Legend, Source

  3. Complete Data Source Inventory

  4. Satellite Data to 30-Metre Analysis Grid — How It Works

  5. Formula Terms — Origin and Scientific Explanation

  6. Derived Indices — Full Methodology, Validation, and Limitations

  7. References

Dashboard Overview

What This Dashboard Is

This is an interactive, web-based geospatial analysis tool built on Google Earth Engine (GEE) — a cloud computing platform operated by Google that provides access to a global catalogue of petabytes of satellite imagery and geospatial datasets. The dashboard is designed specifically for Delhi, India, the world's second-most populous megacity, which regularly experiences severe and lethal heat events during the pre-monsoon and monsoon seasons.

The user selects any calendar year between 2015 and 2026, and any month, then clicks Run. The system automatically:

  1. Downloads and processes Landsat 8/9 satellite images to measure how hot the ground surface is (Land Surface Temperature, LST) at 30-metre pixel resolution.

  2. Downloads and processes Sentinel-2 satellite images (or falls back to Landsat when Sentinel-2 is unavailable) to measure vegetation cover (NDVI), built-up density (NDBI), and water body presence (MNDWI) for that month.

  3. Downloads ERA5-Land hourly reanalysis data from the European Centre for Medium-Range Weather Forecasts (ECMWF) to extract monthly average 2-metre air temperature, dewpoint temperature, 10-metre wind speed, and downwelling solar radiation for the Delhi region.

  4. Applies a statistical downscaling model to translate the coarse ERA5 meteorology (grid spacing approximately 9–11 km) down to a nominal 30-metre spatial resolution, using the satellite-observed surface patterns as spatial anchors.

  5. Computes three internationally recognised thermal comfort and physiological heat stress indices — WBGT, UTCI, and Heat Index (HI) — at 30-metre resolution.

  6. Computes a Heat Vulnerability Index (HVI) using four independent weighting methods, then automatically selects the one that best tracks the observed surface temperature pattern.

  7. Displays all outputs as interactive maps with dynamic, month-specific legends, a pixel inspector (click any location to read values), ward-level summaries for all Delhi wards, and a year-vs-year comparison tool covering 2015 to 2025.

Why This Dashboard Was Built — The Core Problem

India's capital region experiences some of the most extreme heat events in South Asia. Between 2010 and 2022, Indian Heat Action Plans documented thousands of excess deaths attributable to heat. Yet heat is not uniform across a city. In Delhi, a ward covered in dense asphalt and concrete rooftops can be 8–12 °C hotter at the surface than a ward with park cover or a water body, even when both wards are reported as the same temperature by a single weather station.

Conventional approaches to heat monitoring in India rely on a small network of weather stations operated by the India Meteorological Department (IMD) and the Central Pollution Control Board (CPCB). These stations provide accurate point measurements but cannot resolve the sharp spatial gradients within a ward, between a park and a road 500 metres away, or between a slum and a commercial area sharing the same postcode.

This dashboard was built to solve that problem by fusing three independent data streams — satellite surface observations, atmospheric reanalysis, and ground station climatology — to produce spatially continuous heat stress maps at 30-metre resolution, covering all 272+ wards of Delhi for any month from 2015 onwards.

Where This Tool Is Useful

The tool can be used in the following sectors:

  1. Urban and Disaster Management: Identify which specific wards consistently rank highest on LST, air temperature, WBGT, or HVI, as direct inputs to ward-level heat action planning under NDMA guidelines.

  2. Public Health and Epidemiology: Spatially correlate heat stress estimates with health outcome data (hospital admissions, mortality records) at ward scale, enabling exposure-response research in Indian cities.

  3. Environmental and Urban Planning: Quantify the thermal benefit of increasing green cover (NDVI) or reducing built-up density (NDBI) within specific wards; compare pre- and post-intervention months.

  4. Climate Research and Education: Demonstrate multi-year temporal trends in heat indices for Delhi (2015–2025), examine inter-annual variability driven by monsoon timing, urban expansion, or ENSO.

  5. Journalism and Policy Communication: Access a standardised, reproducible heat risk estimate for any ward in any month, rather than relying on anecdotal reports or station averages.