How Remote Sensing and GIS Transform Population Studies

Counting people has always been one of humanity’s most ambitious projects. From village registers to the decennial Census, demographers have spent centuries trying to answer a deceptively simple question: who lives where, and how is that changing? Today, two technologies, Remote Sensing (RS) and Geographic Information Systems (GIS), are quietly rewriting the rules of how population is studied. They let researchers see slums emerge in satellite pixels, track migration through nighttime lights, and predict the next overcrowded suburb before the first brick is laid.

Why population studies needed a new lens 🔗

Traditional population data in India relies heavily on the Census, the Sample Registration System (SRS) and the National Family Health Survey (NFHS). These sources are rich in detail but slow to update; the most recent full Census was conducted in 2011, and the next round has been repeatedly delayed. In a country where cities can double in built-up area within a decade, a ten-year gap is a long time to be blind.

This is where geospatial technologies step in. Remote Sensing captures information about the Earth’s surface using sensors on satellites or aircraft, while GIS stores, analyses and visualises that spatial data alongside demographic, social and environmental layers. Together, they offer something the Census cannot: a near-continuous, spatially detailed view of how human settlements expand, shrink and transform. The National Remote Sensing Centre (NRSC) under ISRO has been a key driver of this shift in India, providing satellite data and platforms that support population and planning research.

Mapping and estimation techniques 🔗

The bridge between satellite pixels and people is built through a set of well-established techniques. The most foundational among them is land use and land cover (LULC) mapping, which classifies every pixel of a satellite image into categories such as built-up area, agricultural land, water bodies, forest and barren land. Once researchers know where built-up areas are and how dense they are, they can begin to estimate population.

Land use and land cover classification 🔗

Satellites like the Indian Remote Sensing series (IRS), Resourcesat, Cartosat and the global Landsat and Sentinel missions capture multispectral imagery at varying resolutions. Using supervised classification methods such as the Maximum Likelihood Classifier, analysts identify built-up areas, vegetation and water with reasonably high accuracy. A study of the National Capital Territory of Delhi reported overall classification accuracy of around 89% with Kappa coefficients above 0.86, which is the kind of reliability planners need before acting on the data.

Population estimation from satellite data 🔗

Once built-up area is mapped, population can be estimated using several approaches. The simplest is the areal interpolation method, where Census population is redistributed across built-up pixels rather than administrative units. More sophisticated methods use building footprints, roof types, and even nighttime lights from satellites like the DMSP-OLS and the newer VIIRS. Researchers have used temporal DMSP-OLS nighttime satellite data to assess urbanisation patterns across India, showing how lit areas can act as a proxy for population concentration.

The advantage of these techniques is granularity. Census data is typically available at the ward or village level, but satellite-derived population grids can produce estimates at 100 m or even 30 m resolution. For health researchers studying disease clusters, sanitation gaps or maternal health access, this level of detail can make the difference between a meaningful intervention and a misdirected one.

Urban growth monitoring 🔗

Time-series satellite data lets researchers watch cities grow almost like a time-lapse film. Comparing imagery from 1990, 2000, 2010 and 2020 reveals not just how much land was urbanised, but where, in which direction, and at what pace. Indices like the Normalised Difference Built-up Index (NDBI) and Shannon’s Entropy are commonly used to quantify the compactness or dispersal of urban form.

Case studies in India 🔗

India has been one of the most studied geographies in geospatial population research, largely because its scale of urbanisation is unprecedented and its satellite programme is robust.

Delhi: a four-decade story of urban sprawl 🔗

The Delhi metropolitan region is perhaps the most analysed example. A long-term study of urban sprawl in Delhi from 1977 to 2014 using Landsat and IRS satellite data documented how the capital’s built-up area expanded outward in waves, with the most aggressive growth occurring along major transport corridors. Earlier work using Landsat MSS, Thematic Mapper and toposheets traced the same expansion from 1975 to 1988, showing how visual interpretation of satellite imagery combined with field checks can map urban sprawl that ground surveys would miss.

A more recent assessment of land use and land cover change in the NCT of Delhi between 2010 and 2021 found significant gains in built-up area at the cost of agricultural and vegetative cover, with the change detection model achieving high Kappa-based accuracy. Studies that combined LISS-III imagery with Landsat data detected informal settlements, fast-growing areas, declining open spaces and shrinking water bodies across the metropolitan region, capturing trends invisible to administrative records.

Population mapping using census and satellite data 🔗

Beyond Delhi, geospatial tools have been used to disaggregate Census data across small geographies. Researchers studying the sub-mountain Siwalik region of Punjab used GIS to map demographic variables at the village level, building thematic maps of population density, sex ratio and scheduled caste distribution while applying inequality measures like the Lorenz curve and Gini coefficient. This village-level granularity helps administrators identify pockets of demographic stress that state-level averages would simply hide.

Platforms like ISRO’s Bhuvan geoportal have democratised this work. Bhuvan offers IRS imagery, administrative boundary layers, and census-linked thematic data that researchers and students can use without specialised software. Users can generate choropleth maps from State or District Census codes, overlay point data, and download free satellite products from missions like Cartosat, Resourcesat and Oceansat. This kind of open infrastructure is what allows a university researcher in a Tier-2 town to do the same analysis as a planner in a metropolitan agency.

Beyond Delhi: smaller cities under the lens 🔗

The same techniques are now being applied to Tier-2 and Tier-3 cities, where most of India’s future urban growth will occur. Studies of the Sonipat-Kundli urban agglomeration in the Delhi-NCR region used Landsat imagery from 1991 to 2021, combined with the Maximum Likelihood Classifier and Shannon’s Entropy model, to track how economic liberalisation accelerated peripheral sprawl. Similar analyses in Kanpur, Hyderabad and Bengaluru have shown how distinct each city’s growth pattern is, leapfrog development in some, concentric expansion in others.

Impact on urban planning and resource management 🔗

The point of all this mapping is not the maps themselves but the decisions they enable. Geospatial data has become central to how Indian cities are planned, financed and governed.

Informing master plans and infrastructure 🔗

Urban master plans now routinely use satellite-derived LULC maps as a baseline. The Ministry of Housing and Urban Affairs, through programmes like AMRUT (Atal Mission for Rejuvenation and Urban Transformation), has supported GIS-based master planning for hundreds of cities. The aim is to ground infrastructure decisions, water supply, sewerage, roads, in actual spatial patterns of population and built-up area rather than outdated paper maps. The Smart Cities Mission has similarly relied on GIS dashboards to integrate data on transport, utilities, environment and citizen services.

Resource allocation and service delivery 🔗

When population is mapped at fine resolution, public services can be targeted with precision. Schools, primary health centres, anganwadis and ration shops can be located based on actual demand surfaces rather than administrative convenience. GIS-based service-area analysis can identify neighbourhoods where the nearest health facility is too far, or where multiple facilities overlap while others are underserved. For a country attempting to deliver universal health coverage, this kind of spatial targeting is invaluable.

Disaster preparedness and environmental health 🔗

Population mapping also feeds directly into disaster risk reduction. Flood-prone wards, landslide-vulnerable hillsides and earthquake-exposed corridors can be overlaid with population grids to estimate how many people are at risk. Following major floods in Kerala and Assam, agencies have used satellite imagery to assess inundation extent and cross-reference it with settlement maps to plan relief. Air quality monitoring stations, dengue cluster mapping and heat-island studies all draw on the same geospatial foundation.

Tracking informal settlements 🔗

One of the most powerful uses of high-resolution satellite imagery is the identification of informal settlements, which are notoriously under-counted in official surveys. By detecting characteristic patterns of dense, low-rise, irregular roof structures, analysts can map slum-like settlements that may not appear in municipal records. This has obvious implications for housing policy, sanitation programmes like the Swachh Bharat Mission, and family-health interventions that need to reach the most marginalised.

Limitations to keep in mind 🔗

Geospatial technologies are powerful but not infallible. Cloud cover can obscure imagery, particularly during the monsoon. Classification accuracy depends on the quality of training data and the analyst’s judgment. Built-up area is not the same as population, two pixels of the same density may house very different numbers of people depending on building height, household size and economic status. And while satellites can count rooftops, they cannot directly measure income, caste, gender or health outcomes, which is why integration with the Census, NFHS and other surveys remains essential.

There are also ethical concerns. As mapping becomes more granular, questions of privacy, surveillance and the right to be uncounted become real. A balance between visibility-for-services and protection-from-surveillance is something planners and policymakers are only beginning to confront.

The road ahead 🔗

The next frontier is the integration of geospatial data with mobile phone records, social media, and Internet of Things sensors to create dynamic, near-real-time population maps. Machine learning models trained on satellite imagery are already producing population estimates at 100 m resolution across the entire country. As India prepares for its next Census and continues to urbanise at one of the fastest rates in the world, remote sensing and GIS will move from being supportive research tools to being core infrastructure of population studies.

What do you think? If you could overlay just one additional dataset on a satellite-derived population map of your city, would it be income, health access, or environmental quality, and what would that combined map reveal that the Census never could?

References

1. https://www.nrsc.gov.in/
2. https://link.springer.com/article/10.1007/s43538-023-00152-2
3. https://www.currentscience.ac.in/Volumes/100/10/1479.pdf
4. https://journals.ametsoc.org/view/journals/eint/20/14/ei-d-15-0040.1.xml
5. https://link.springer.com/article/10.1007/BF02995831
6. https://spie.org/news/0910-urban-remote-sensing-for-a-fast-growing-megacity-delhi-india
7. https://link.springer.com/article/10.1007/BF03013489
8. https://bhuvan.nrsc.gov.in/
9. https://amrut.mohua.gov.in/
10. https://smartcities.gov.in/
11. https://swachhbharatmission.ddws.gov.in/