Land use regression model

A land use regression model (LUR model) is an algorithm often used for analyzing pollution, particularly in densely populated areas.

The model is based on predictable pollution patterns to estimate concentrations in a particular area. This requires some linkage to the environmental characteristics of the area, especially characteristics that influence pollutant emission intensity and dispersion efficiency. LUR modeling is a useful approach for screening studies and can substitute for dispersion models given insufficient input data or dispersion models.

Multiple regression equations are used to describe the relationship between sample locations and environmental variables, often relying on geographic information systems (GIS) to collect measurements. This results in an equation that can predict pollution concentrations at unmeasured locations based on data for the predictor variables in specific locations. A raster graphic image of the area is generated and intersected with area-level population data to formulate the exposure distribution.

Application examples

edit

Health impact assessments

edit

LUR models were originally developed to assess the exposure resulting from air pollution as a result of vehicular traffic, but they have since been expanded to cover air pollution epidemiology. The EPA has an ongoing grant for these types of assessments, where they collect hourly updates across three major U.S. cities to study how pollution concentrations change over time and track health effects reported by those who live there.[1]

Liaoning Province, China

edit

A study incorporating annual satellite aerosol optical depth (AOD) observations and five specific canyon indicators (building height, coverage ratio, shape coefficient, floor-area ration, and skyscraper-building ratio) was used to successfully enhance the LUR's modeling accuracy. The area picked for this study had been affected by rapid urbanization which resulted in serious environmental atmospheric pollution and measured a few key pollutants, in particular (PM2.5, PM10, SO2, NO2, NOx, CO, and O3).

Ontario, Canada

edit

LUR was used to predict the concentration of benzene, toluene, ethylbenzene, m/p-xylene, and o-xylene (BTEX) concentrations in Ontario. 39 locations were monitored for 2 weeks to support LUR models to have predictor variables and best estimate BTEX concentrations. [2]

Gothenburg, Sweden

edit

A 20-year study examined urban pollution in Gothenburg (an urban area in Sweden) about the NO2 concentration. The results were accurate in the effects of altitude and traffic intensity on pollution in a certain region. The model was used to estimate outdoor concentrations in urban areas, but not accurate in less populated regions, such as islands or rural areas.[3]

Further development

edit

LUR can be expanded to encapsulate less studied areas with similar characteristics.

Mobile monitoring

edit

Mobile monitoring is an alternative to traditional fixed-site measuring systems. Mobile monitoring enables good spatial coverage even with few monitoring devices. Mobile monitoring enables investigations given monetary constraints.[4]

Additional prediction variables

edit

Adding prediction variables can increase the accuracy and/or explanatory power of LUR models. Examples include holidays traffic variants and seasonal meteorological variations.

Geographically weighted models

edit

The incorporation of Geographically Weighted Regression (GWR) into LURs involves applying a spatial weighting function to the spatial coordinates that divide a study area into various local neighborhoods. This can reduce the effects of spatial non-stationarity, a defect that occurs when variables form inconsistent relationships over large areas, misrepresenting data points as unchanging.[5]

Alternatives

edit

Alternatives to LUR include kriging, atmospheric dispersion modeling and Bayesian Maximum Entropy modeling.[6]

References

edit
  1. ^ Robinson, Allen. "Next Generation LUR Models: Development of Nationwide Modeling Tools for Exposure Assessment and Epidemiology". EPA. Retrieved 2018-11-26.
  2. ^ Atari, Dominic (2009). "Assessing the distribution of volatile organic compounds using land use regression in Sarnia, Ontario, Canada". Environmental Health. 8: 16. doi:10.1186/1476-069X-8-16. PMC 2679013. PMID 19371421.
  3. ^ Habermann, Mateus (2015). "Land use Regression as Method to Model Air Pollution. Previous Results for Gothenburg/Sweden". Procedia Engineering. 115: 21–28. doi:10.1016/j.proeng.2015.07.350.
  4. ^ Hankey, Steve (2015). "Land Use Regression Models of On-Road Particulate Air Pollution (Particle Number, Black Carbon, PM2.5, Particle Size) Using Mobile Monitoring". Environmental Science & Technology. 49 (15): 9194–9202. Bibcode:2015EnST...49.9194H. doi:10.1021/acs.est.5b01209. PMID 26134458.
  5. ^ Bertazzon, Stefania (2015). "Accounting for spatial effects in land use regression for urban air pollution modeling". Spatial and Spatio-Temporal Epidemiology. 14–15: 9–21. doi:10.1016/j.sste.2015.06.002. PMID 26530819.
  6. ^ Adam-Poupart, Ariane (2014). "Spatiotemporal Modeling of Ozone Levels in Quebec (Canada): A Comparison of Kriging, Land-Use Regression (LUR), and Combined Bayesian Maximum Entropy–LUR Approaches". Environmental Health Perspectives. 122 (9): 970–976. doi:10.1289/ehp.1306566. PMC 4153742. PMID 24879650.