Land Use Regression Models for Air Quality Assessment

Land use regression (LUR) models have emerged as a cornerstone of modern exposure assessment, offering finely resolved estimates of ambient pollutant concentrations across urban and regional landscapes. By relating measured levels of pollutants such as fine particulate matter (PM₂.₅), nitrogen dioxide (NO₂) and ultrafine particles to spatial predictors—including traffic intensity, land cover, population density and topography—these models capture local gradients in pollution that are not resolved by regulatory monitoring networks alone. Recent advances have extended conventional LUR frameworks by incorporating spatiotemporal components, low-cost sensor networks, mobile monitoring campaigns and satellite observations, thereby enabling continuous mapping over both space and time. The integration of advanced statistical and machine learning techniques—ranging from partial least squares regression and random forest algorithms to hybrid geostatistical approaches—has enhanced predictive performance and facilitated model transferability across study areas. Such developments underpin epidemiological investigations of long-term health effects, guide urban planning and inform policy interventions aimed at mitigating exposure disparities and improving air quality at the population level.

Research from Nature Portfolio

Recent work has foregrounded the interplay between novel exposure assessment methods and policy action by harnessing granular measurements of air pollutants alongside individual-level data on economic, behavioural and environmental variables. This research has demonstrated how targeted local policies—such as traffic restrictions and infrastructure modifications—can be evaluated against high-resolution exposure gradients, thereby closing the loop between measurement science and practical interventions to reduce health burdens in vulnerable communities.

Another study undertook a multidimensional performance comparison of land use regression and ordinary kriging interpolation for mapping particulate matter concentrations. By evaluating both point-based error metrics and area-based statistics such as information entropy, it showed that LUR models produce more detailed spatial variation than kriging alone, while remaining competitive in overall accuracy. The work emphasises the value of integrating point- and area-based validation to capture nuanced patterns in urban pollution fields.

Land Use Regression Models for Air Quality Assessment publication trend

The graph below shows the total number of articles in land use regression models for air quality assessment across all publications each year (not limited to Nature Index journals).

Technical terms

Land use regression (LUR): A statistical modelling approach that quantifies the relationship between measured pollutant concentrations and spatial predictor variables to estimate air quality across unsampled locations.

Spatiotemporal modelling: Techniques that account for both spatial and temporal variation in pollutant levels by integrating data streams over time and across geographic space.

Ordinary kriging (OK): A geostatistical interpolation method that predicts values at unmonitored sites using weighted averages of nearby observations, based on spatial autocorrelation.

Integrated empirical geographic (IEG) regression: A hybrid modelling framework combining empirical regression of geographic covariates with spatial interpolation methods, such as kriging, to predict pollutant concentrations.

Cross-validation: A model evaluation technique in which data are partitioned into training and testing subsets to assess predictive accuracy and guard against overfitting.

References

1. A comprehensive review of the development of land use regression approaches for modeling spatiotemporal variations of ambient air pollution: A perspective from 2011 to 2023. Environment International (2024).
2. A comparison of linear regression, regularization, and machine learning algorithms to develop Europe-wide spatial models of fine particles and nitrogen dioxide. Environment International (2019).
3. Concentrations of criteria pollutants in the contiguous U.S., 1979 – 2015: Role of prediction model parsimony in integrated empirical geographic regression. PLOS ONE (2020).
4. Advancing environmental exposure assessment science to benefit society. Nature Communications (2019).
5. Performance comparison of LUR and OK in PM2.5 concentration mapping: a multidimensional perspective. Scientific Reports (2015).