Low-cost sensors and microscale land use regression: Data fusion to resolve air quality variations with high spatial and temporal resolution

Low-cost sensor networks and land-use regression: Interpolating nitrogen dioxide concentration at high temporal and spatial resolution in Southern California

“…While regulatory air quality monitoring networks offer important insights about long-term air quality trends, the data must be supplemented with additional measurements and models to obtain geographically more detailed air pollution information (Li et al, 2019a). This is of importance given that air pollutants vary considerably over small distances (Kumar et al, 2015;Weissert et al, 2019a).…”

“…If deployed in dense networks, low-cost sensors have the potential to provide near real-time measurements of pollutants at a spatial resolution representative of the neighbourhood scale. They can offer insights into the influence of local pollution sources at different temporal and spatial scales that may not be detected by the usually sparsely distributed regulatory monitoring networks (Feinberg et al, 2019;Li et al, 2019b;Popoola et al, 2018;Weissert et al, 2019a). Hence, the increasingly available data from lowcost sensor networks has led to new research aimed at combining continuous measurements obtained from a low-cost sensor network with land use data to get spatially and temporally dense air pollution information (Deville Cavellin et al, 2016;Lim et al, 2019;Masiol et al, 2019;Miskell et al, 2018b;Schneider et al, 2017).…”

“…A network of ten low-cost sensors was used in New York to develop 24 LUR models representative of each hour of the day (Masiol et al, 2019). In a recently published pilot study, we presented another approach to combine NO2 concentrations obtained from a microscale low-cost sensor network (eight sensors) with land use information to identify site and time specific effects of urban design features that disproportionately contribute to population exposure (Weissert et al, 2019a).…”

“…In our recent work, we developed calibration and remote drift detection procedures for O3 and NO2 sensors deployed in hierarchical networks consisting of well-maintained regulatory sites and low-cost sensors. The approaches were successfully applied and tested at co-located sites (Miskell et al, 2019;Miskell et al, 2016;Miskell et al, 2018a;Weissert et al, 2019a) and were extended to a larger network in Southern California. The corrected sensor data provide a spatially and temporally dense data set of NO2 and O3 concentrations that can claim reliability with an root mean-square error (RMSE) of 7.4 and 5.4 ppb, respectively.…”

“…The corrected sensor data provide a spatially and temporally dense data set of NO2 and O3 concentrations that can claim reliability with an root mean-square error (RMSE) of 7.4 and 5.4 ppb, respectively. Here, first we use this dataset to develop a land-use model for concentrations averaged over two months, and then apply the simple approach described by Weissert et al (2019a) to model concentrations on an hourly time-scale, both for NO2 and O3. Subsequently, we use an analysis of differences between measured and modelled concentrations to identify local urban conditions that are poorly captured by the static land use model.…”