Prediction of ultrafine particle number concentrations in urban environments by means of Gaussian process regression based on measurements of oxides of nitrogen

Reggente, Matteo; Peters, Jan; Theunis, Jan; Poppel, Martine Van; Rademaker, Michaël; Kumar, Prashant; Baets, Bernard De

doi:10.1016/j.envsoft.2014.07.012

Cited by 27 publications

(17 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hegstad and Omre, 2001;Craig et al, 2001), atmospheric dispersion (e.g. Politis and Robertson, 2004;Konda et al, 2010;Reggente et al, 2014) and climatology (Rougier et al, 2009b;Qin et al, 2013;Castruccio et al, 2014;Plouffe et al, 2015). In contrast, statistical emulation is relatively unknown to the computational wind engineering (CWE) community.…”

Section: Introductionmentioning

confidence: 96%

Employing statistical model emulation as a surrogate for CFD

Moonen

Allegrini

2015

Environmental Modelling & Software

View full text Add to dashboard Cite

Section: Introductionmentioning

confidence: 96%

Employing statistical model emulation as a surrogate for CFD

Moonen

Allegrini

2015

Environmental Modelling & Software

View full text Add to dashboard Cite

“…Gaussian processes (GP) (Williams and Rasmussen (1996); Rasmussen (1997); Rasmussen and Williams (2005)) are nonparametric statistical models that compactly describe distributions over functions with continuous do-mains. They have found various applications in the environmental modeling community, where they are used as data-driven models capable to predict various quantities of interest with quantified uncertainties such as ultra fine particles (Reggente et al (2014)), mean temperatures over North Atlantic Ocean (Higdon (1998)), wind speed (Hu and Wang (2015)), and monthly streamflow (Sun et al (2014)), just to name a few. When the training data for GPs comes from simulators rather than field measurements, then GPs become computational efficient surrogate models or emulators of highfidelity models (Kennedy et al (2002); O'Hagan (2006); Conti and O'Hagan (2010)), with various applications in environmental modeling such as fire emissions (Katurji et al (2015)), ocean and climate circulation (Tokmakian et al (2012)), urban drainage (Machac et al (2016)), and computational fluid dynamics (Moonen and Allegrini (2015)).…”

Section: Introductionmentioning

confidence: 99%

Geospatial uncertainty modeling using Stacked Gaussian Processes

Abdelfatah

Bao

Terejanu

2018

Environmental Modelling & Software

View full text Add to dashboard Cite

“…To obtain better prediction performance, Clifford et al [22] proposed a generalized additive model using meteorological data, time, solar radiation and rainfall as explanatory variables. Reggente et al [23] employed a Gaussian process regression to estimate UFPs in an urban air pollution monitoring network based on local and remote concentrations of NO x , O 3 , CO, and UFPs. None of the mentioned works have considered mobile sensor networks.…”

Section: B Air Pollution Modelingmentioning

confidence: 99%

High Resolution Air Pollution Maps in Urban Environments Using Mobile Sensor Networks

Marjovi

Arfire

Martinoli

2015

2015 International Conference on Distributed Computing in Sensor Systems

View full text Add to dashboard Cite

Abstract-We propose three modeling methods using a mobile sensor network to generate high spatio-temporal resolution air pollution maps for urban environments. In our deployment in Lausanne (Switzerland), dedicated sensing nodes are anchored to the public buses and measure multiple air quality parameters including the Lung Deposited Surface Area (LDSA), a state of the art metric for quantifying human exposure to ultrafine particles. In this paper, our focus is on generating LDSA maps. In particular, since the sensor network coverage is spatially and temporally dynamic, we leverage models to estimate the values for the locations and times where the data are not available. We first discretize the area topologically based on the street segments in the city and we then propose the following three prediction models: i) a log-linear regression model based on nine meteorological (e.g., temperature and precipitations) and gaseous (e.g., NO2 and CO) explanatory variables measured at two static stations in the city, ii) a novel network-based log-linear regression model that takes into account the LDSA values of the most correlated streets and also the nine explanatory variables mentioned above, iii) a novel Probabilistic Graphical Model (PGM) in which each street segment is considered as one node of the graph, and inference on conditional joint probability distributions of the nodes results in estimating the values in the nodes of interest. More than 44 millions of geo-and time-stamped LDSA measurements (i.e., more than 14 months of real data) are used in this paper to evaluate the proposed modeling approaches in various time resolutions (hourly, daily, weekly and monthly). The results show that the three approaches bring significant improvements in R 2 , RMSE and FAC metrics compared to a baseline KNearest Neighbor method.

show abstract

Prediction of ultrafine particle number concentrations in urban environments by means of Gaussian process regression based on measurements of oxides of nitrogen

Cited by 27 publications

References 47 publications

Employing statistical model emulation as a surrogate for CFD

Employing statistical model emulation as a surrogate for CFD

Geospatial uncertainty modeling using Stacked Gaussian Processes

High Resolution Air Pollution Maps in Urban Environments Using Mobile Sensor Networks

Contact Info

Product

Resources

About