%0 Journal Article %1 su2015modeling %A Su, Jason G. %A Hopke, Philip K. %A Tian, Yilin %A Baldwin, Nichole %A Thurston, Sally W. %A Evans, Kristin %A Rich, David Q. %D 2015 %J Atmospheric Environment %K lursurvey %P 477 - 483 %R https://doi.org/10.1016/j.atmosenv.2015.10.002 %T Modeling particulate matter concentrations measured through mobile monitoring in a deletion/substitution/addition approach %U http://www.sciencedirect.com/science/article/pii/S1352231015304210 %V 122 %X Land use regression modeling (LUR) through local scale circular modeling domains has been used to predict traffic-related air pollution such as nitrogen oxides (NOX). LUR modeling for fine particulate matters (PM), which generally have smaller spatial gradients than NOX, has been typically applied for studies involving multiple study regions. To increase the spatial coverage for fine PM and key constituent concentrations, we designed a mobile monitoring network in Monroe County, New York to measure pollutant concentrations of black carbon (BC, wavelength at 880 nm), ultraviolet black carbon (UVBC, wavelength at 3700 nm) and Delta-C (the difference between the UVBC and BC concentrations) using the Clarkson University Mobile Air Pollution Monitoring Laboratory (MAPL). A Deletion/Substitution/Addition (D/S/A) algorithm was conducted, which used circular buffers as a basis for statistics. The algorithm maximizes the prediction accuracy for locations without measurements using the V-fold cross-validation technique, and it reduces overfitting compared to other approaches. We found that the D/S/A LUR modeling approach could achieve good results, with prediction powers of 60%, 63%, and 61%, respectively, for BC, UVBC, and Delta-C. The advantage of mobile monitoring is that it can monitor pollutant concentrations at hundreds of spatial points in a region, rather than the typical less than 100 points from a fixed site saturation monitoring network. This research indicates that a mobile saturation sampling network, when combined with proper modeling techniques, can uncover small area variations (e.g., 10 m) in particulate matter concentrations.

@article{su2015modeling, abstract = {Land use regression modeling (LUR) through local scale circular modeling domains has been used to predict traffic-related air pollution such as nitrogen oxides (NOX). LUR modeling for fine particulate matters (PM), which generally have smaller spatial gradients than NOX, has been typically applied for studies involving multiple study regions. To increase the spatial coverage for fine PM and key constituent concentrations, we designed a mobile monitoring network in Monroe County, New York to measure pollutant concentrations of black carbon (BC, wavelength at 880 nm), ultraviolet black carbon (UVBC, wavelength at 3700 nm) and Delta-C (the difference between the UVBC and BC concentrations) using the Clarkson University Mobile Air Pollution Monitoring Laboratory (MAPL). A Deletion/Substitution/Addition (D/S/A) algorithm was conducted, which used circular buffers as a basis for statistics. The algorithm maximizes the prediction accuracy for locations without measurements using the V-fold cross-validation technique, and it reduces overfitting compared to other approaches. We found that the D/S/A LUR modeling approach could achieve good results, with prediction powers of 60%, 63%, and 61%, respectively, for BC, UVBC, and Delta-C. The advantage of mobile monitoring is that it can monitor pollutant concentrations at hundreds of spatial points in a region, rather than the typical less than 100 points from a fixed site saturation monitoring network. This research indicates that a mobile saturation sampling network, when combined with proper modeling techniques, can uncover small area variations (e.g., 10 m) in particulate matter concentrations.}, added-at = {2020-01-15T17:41:28.000+0100}, author = {Su, Jason G. and Hopke, Philip K. and Tian, Yilin and Baldwin, Nichole and Thurston, Sally W. and Evans, Kristin and Rich, David Q.}, biburl = {https://www.bibsonomy.org/bibtex/22717a3128c04ca13e6dc6477e9bcd413/lautenschlager}, description = {Modeling particulate matter concentrations measured through mobile monitoring in a deletion/substitution/addition approach - ScienceDirect}, doi = {https://doi.org/10.1016/j.atmosenv.2015.10.002}, interhash = {b72de18f5a9b8f1c7f3a9220ac3d7b7b}, intrahash = {2717a3128c04ca13e6dc6477e9bcd413}, issn = {1352-2310}, journal = {Atmospheric Environment}, keywords = {lursurvey}, pages = {477 - 483}, timestamp = {2020-01-15T17:41:28.000+0100}, title = {Modeling particulate matter concentrations measured through mobile monitoring in a deletion/substitution/addition approach}, url = {http://www.sciencedirect.com/science/article/pii/S1352231015304210}, volume = 122, year = 2015 }