Delhi’s urban form shapes PM2.5 differently across scales, study finds

A multiscale spatial analysis found that the relationship between Delhi’s urban form and fine-particle pollution changed across the city and between four study years. Using satellite-derived PM2.5 estimates for 2010, 2015, 2020 and 2023, the research combined urban-form metrics with environmental, meteorological and socioeconomic variables in global and geographically weighted regression models. The multiscale model explained more of the observed variation than the global model, with adjusted R-squared values ranging from 0.825 to 0.908 across the selected years.

The study examined PM2.5 at approximately 1-kilometre resolution, allowing the analysis to compare pollution patterns across Delhi rather than treating the city as a single uniform environment. The researchers used land-use and land-cover data to derive metrics of fragmentation, sprawl, compactness and diversity through FRAGSTATS. These morphological measures were assessed alongside vegetation index, precipitation, temperature, wind speed, population density and GDP per capita. The resulting comparison tested whether the same urban characteristics had consistent effects throughout the metropolitan area or operated differently from one location to another.

Multiscale Geographically Weighted Regression, or MGWR, outperformed the global regression model in explaining the spatial variation in PM2.5. Its stronger performance indicates that relationships within the city are not adequately represented by one citywide coefficient applied everywhere. Instead, the importance and direction of urban-form, environmental, meteorological and socioeconomic relationships can vary across space and at different spatial scales. The abstract does not provide the coefficient values for individual predictors, so it does not establish which particular urban-form metric increased or decreased PM2.5 at a given location.

The study also found that the urban-form metrics associated with PM2.5 differed among 2010, 2015, 2020 and 2023, suggesting that the morphological characteristics of the city and their relationships with particle pollution were not fixed over the 13-year period. Environmental and meteorological variables generally had stronger and more spatially heterogeneous effects at finer operational scales. By contrast, urban-form and socioeconomic relationships were described as weaker but broader and more spatially stable. This contrast means that citywide planning variables may persist across larger areas even as local conditions and the operative scales of influence change.

The results do not demonstrate that altering urban form alone would reduce Delhi’s PM2.5. The analysis is observational and depends on satellite-derived pollution estimates, land-cover data and statistical associations; it does not isolate the causal effect of a particular development pattern or planning intervention. The study is also limited to the selected years, the variables included in the models and the spatial resolution available for the data. Nevertheless, its adjusted R-squared range and the advantage of MGWR over a global model support the use of multiscale spatial methods when analysing pollution in a large, internally varied city. The findings point to an air-quality strategy that evaluates urban form together with green infrastructure, climate-responsive design and local environmental conditions rather than relying only on metropolitan averages.

Subject of Research: Multiscale analysis of urban morphology and PM2.5 variation in Delhi

Article Title: Multiscale spatial assessment of urban morphology on PM2.5 dynamics in Delhi using MGWR model

Article References: Sen, H., & Gupta, L. (2026). Multiscale spatial assessment of urban morphology on PM2.5 dynamics in Delhi using MGWR model. Air Quality, Atmosphere & Health, 19(11), Article 244. https://doi.org/10.1007/s11869-026-02131-0

Image Credits: AI Generated

DOI: 10.1007/s11869-026-02131-0

Keywords: Delhi, PM2.5, urban morphology, MGWR, air pollution, satellite data, spatial modelling, land use, meteorology, urban planning, Multiscale, spatial