Land Use Regression Models for Ultrafine Particles and Black Carbon Based on Short-Term Monitoring Predict Past Spatial Variation

Abstract

Health effects of long-term exposure to ultrafine particles (UFP) have not been investigated in epidemiological studies because of the lack of spatially resolved UFP exposure data. Short-term monitoring campaigns used to develop land use regression (LUR) models for UFP typically had moderate performance. The aim of this study was to develop and evaluate spatial and spatiotemporal LUR models for UFP and Black Carbon (BC), including their ability to predict past spatial contrasts. We measured 30 min at each of 81 sites in Amsterdam and 80 in Rotterdam, The Netherlands in three different seasons. Models were developed using traffic, land use, reference site measurements, routinely measured pollutants and weather data. The percentage explained variation (R(2)) was 0.35-0.40 for BC and 0.33-0.42 for UFP spatial models. Traffic variables were present in every model. The coefficients for the spatial predictors were similar in spatial and spatiotemporal models. The BC LUR model explained 61% of the spatial variation in a previous campaign with longer sampling duration, better than the model R(2). The UFP LUR model explained 36% of UFP spatial variation measured 10 years earlier, similar to the model R(2). Short-term monitoring campaigns may be an efficient tool to develop LUR models.

Land Use Regression Models for Ultrafine Particles in Six European Areas

“…12,24 First, the mean reference UFP concentration of the corresponding interval was subtracted from the annual mean concentration at the reference site. Second, this difference was added to the concentration measured at a site.…”

“…7,12 Briefly, temporal-variation adjusted 30 min average UFP concentration per site was used as dependent variable in a linear regression model, using GIS predictors as explanatory variables. Predictors where the 90th percentile was zero were not used in any model.…”

“…Due to correlations between predictor variables, it is likely that alternative models can be developed which explain variability almost equally well. 12 Gulliver 19 developed and interpreted four NO 2 models in the framework of 4-fold hold-out validation (HV). Wang 20 applied model predictions of 40 models from a cross-validation method to predict subject’s exposure to NO 2 in an epidemiological study.…”

Development and evaluation of land use regression models for black carbon based on bicycle and pedestrian measurements in the urban environment

“…Some other studies use hold-out validation (HV) or random n-fold cross-validation to get a more reliable estimate of the performance (e.g. Hasenfratz et al, 2015;Kanaroglou et al, 2013;Montagne et al, 2015). For these approaches, sufficient sampling locations are required to be able to split up the dataset.…”

“…Some studies use a mobile platform to perform short-term measurements at many locations (e.g. Larson et al, 2009;Merbitz et al, 2012;Ghassoun et al, 2015;Montagne et al, 2015). Only few studies use mobile measurements as a basis for LUR modelling.…”