Prediction of Nitrogen Dioxide (NO2) on Ground Level using Land Use Regression and Ordinary Kriging with Anisotropy

Authors: Changyeon Lee*, College of Design, Construction and Planning
Topics: Environmental Science, United States, Geographic Information Science and Systems
Keywords: NO2 prediction, Land Use Regression, Ordinary Krigin, United States,
Session Type: Paper
Day: 4/14/2018
Start / End Time: 8:00 AM / 9:40 AM
Room: Napoleon B1, Sheraton 3rd Floor
Presentation File: No File Uploaded


Measurement of air pollution on ground level is based on monitoring stations as point sources because the cost constructing and operating each station is so expensive. Therefore, many studies predict air pollution level on ground level. As a typical method, land use regression (LUR) is used. Recently, many studies use hybrid models which combine LUR and other methods. This study predicts NO2 concentration on ground level over U.S. territory in 2014. This study not only uses LUR, but also ordinary kriging (OK) with anisotropy with residuals in LUR. LUR is a statistic method based on multivariate linear regressions, this study mainly uses partial least square (PLS) to remove multicollinearity between variables. This study separate independent variables to five factors: emission sources, diffusing/concentration sources, chemical reaction, binary variable, and ozone monitoring instrument (OMI) measured by AURA satellite. As main emission sources, there are vehicle roads, railroads, ports, and airports. Diffusing/concentration sources are related to regional environments affecting diffusion or concentration of NO2: green area, elevation, precipitation, wind speed, and distance from coastlines. Temperature and solar insolation are related to chemical reaction of air pollutants. As binary variables, there are urban or non-urban area, and three regions over U.S. like west, mountain, and east. OMI provide the another powerful new information. As geo-statistical method, OK is one of interpolation method including Inverse Distance Weighting (IDW) and spatial correlation. In addition, using anisotropy in OK means that there is the directional difference in spatial correlation.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login