Data fusion of sparse, heterogeneous, and mobile sensor devices using adaptive distance attention DOI Creative Commons
Jean-Marie Lepioufle, Philipp Schneider, Paul Hamer

et al.

Environmental Data Science, Journal Year: 2024, Volume and Issue: 3

Published: Jan. 1, 2024

Abstract In environmental science, where information from sensor devices are sparse, data fusion for mapping purposes is often based on geostatistical approaches. We propose a methodology called adaptive distance attention that enables us to fuse heterogeneous, and mobile predict values at locations with no previous measurement. The approach allows automatically weighting the measurements according priori quality about device without using complex resource-demanding assimilation techniques. Both ordinary kriging general regression neural network (GRNN) integrated into this their learnable parameters deep learning architectures. evaluate method three static phenomena different complexities: case related simplistic phenomenon, topography over an area of 196 $ {km}^2 annual hourly {NO}_2 concentration in 2019 Oslo metropolitan region (1026 ). simulate networks 100 synthetic six characteristics measurement spatial resolution. Generally, outcomes promising: we significantly improve metrics baseline models. Besides, Nadaraya–Watson kernel provides as good system enabling possibility alleviate processing cost sparse data. encouraging results motivate keeping adapting space-time evolving isolated areas.

Language: Английский

Challenges and opportunities of low-cost sensors in capturing the impacts of construction activities on neighborhood air quality DOI Creative Commons
Weaam Jaafar, Junshi Xu,

Emily Farrar

et al.

Building and Environment, Journal Year: 2024, Volume and Issue: 254, P. 111363 - 111363

Published: March 11, 2024

In large metropolitan areas such as Toronto, planners are increasingly relying on urban densification to accommodate population growth sustainably. While infill developments support the city's long-term climate goals, on-going construction impacts air quality for local communities. Understanding how neighborhoods impacted by these localized sources can be achieved implementing a network of low-cost sensors. this study, we placed twelve sensors balconies in Toronto neighborhood various projects. The study aims capture impact and heavy-duty traffic provide better understanding spatial variability fine particulate matter (PM2.5). locations were compared using time series analysis, inverse distance weighing (IDW) heterogeneity, spectral analysis quantify contribution sources. Sensors exhibited inter-sensor variability, which was corrected upon calibration. located near far from sites showed similar temporal trends, however measured greater PM2.5 concentrations, where hourly average concentration ranged between 6.8 8.5 μg/m3 further away 5.4 6.2 μg/m3. Spatial also captured IDW more heterogenous concentrations. Spectral demonstrated that contributed up 23% levels while distant had maximum 11% contribution. By sensors, explore create hot spots within neighborhood.

Language: Английский

Citations

6
Route selection for real-time air quality monitoring to maximize spatiotemporal coverage DOI
Rashmi Choudhary, Amit Agarwal

Journal of Transport Geography, Journal Year: 2024, Volume and Issue: 115, P. 103812 - 103812

Published: Jan. 30, 2024

Language: Английский

Citations

5
A novel spatiotemporal prediction approach to fill air pollution data gaps using mobile sensors, machine learning and citizen science techniques DOI Creative Commons
Arunik Baruah, Dimitrios Bousiotis, Seny Damayanti

et al.

npj Climate and Atmospheric Science, Journal Year: 2024, Volume and Issue: 7(1)

Published: Dec. 19, 2024

Abstract Particulate Matter (PM) air pollution poses significant threats to public health. We introduce a novel machine learning methodology predict PM 2.5 levels at 30 m long segments along the roads and temporal scale of 10 seconds. A hybrid dataset was curated from an intensive campaign in Selly Oak, Birmingham, UK, utilizing citizen scientists low-cost instruments strategically placed static mobile settings. Spatially resolved proxy variables, meteorological parameters, properties were integrated, enabling fine-grained analysis . Calibration involved three approaches: Standard Random Forest Regression, Sensor Transferability Road Evaluations. This significantly increased spatial resolution beyond what is possible with regulatory monitoring, thereby improving exposure assessments. The findings underscore importance approaches science advancing our understanding pollution, small number participants enhancing local quality assessment for thousands residents.

Language: Английский

Citations

4
Toward causal artificial intelligence approach for PM2.5 interpretation: A discovery of structural causal models DOI Creative Commons
Mallika Kliangkhlao,

Apaporn Tipsavak,

Thanathip Limna

et al.

Ecological Informatics, Journal Year: 2025, Volume and Issue: unknown, P. 103115 - 103115

Published: March 1, 2025

Language: Английский

Citations

0
Application of Proper Orthogonal Decomposition to Elucidate Spatial and Temporal Correlations in Air Pollution Across the City of Liverpool, UK DOI Creative Commons
Cammy Acosta‐Ramírez, Jonathan Higham

Urban Science, Journal Year: 2025, Volume and Issue: 9(5), P. 166 - 166

Published: May 13, 2025

Understanding the spatiotemporal distribution of air pollution is critical for improving urban quality. Advances in wireless sensor networks have made it possible to monitor across cities at higher resolutions. The new spatial coverage allows novel implementation advanced statistical methods detect spatially important, coherent patterns environmental flows. In this study, we apply proper orthogonal decomposition a derived from 34 particulate matter sensors, which collected data over 250 days Liverpool City Region England, identify set modes. dominant mode exhibits daily periodicity increases matter, with residential areas interpreted as changes driven by commutes. second highlights seasonal changes, and third alludes transportation simultaneous decreases. contrast traditional time series analyses, enables elucidation that otherwise might remain hidden. Our findings highlight benefits demonstrate applicability studying movements polluted their correlations meteorological variables anthropogenic factors.

Language: Английский

Citations

0
Assessing the Spatial Transferability of Calibration Models across a Low-cost Sensors Network DOI
Vasudev Malyan, Vikas Kumar,

Mufaddal Moni

et al.

Journal of Aerosol Science, Journal Year: 2024, Volume and Issue: 181, P. 106437 - 106437

Published: July 20, 2024

Language: Английский

Citations

3
Pinpointing sources of pollution using citizen science and hyperlocal low-cost mobile source apportionment DOI Creative Commons
Dimitrios Bousiotis, Seny Damayanti, Arunik Baruah

et al.

Environment International, Journal Year: 2024, Volume and Issue: 193, P. 109069 - 109069

Published: Oct. 11, 2024

Currently, methodologies for the identification and apportionment of air pollution sources are not widely applied due to their high cost. We present a new approach, combining mobile measurements from multiple sensors collected daily walks citizen scientists, in population density area Birmingham, UK. The methodology successfully pinpoints different affecting local quality using only handful measurements. It was found that regional were mostly responsible PM

Language: Английский

Citations

3
Quantifying and predicting air quality on different road types in urban environments using mobile monitoring and automated machine learning DOI
Chunping Miao, Zhong‐Ren Peng,

Aiwei Cui

et al.

Atmospheric Pollution Research, Journal Year: 2023, Volume and Issue: 15(3), P. 102015 - 102015

Published: Dec. 15, 2023

Language: Английский

Citations

8
LSTM Deep Learning Models for Virtual Sensing of Indoor Air Pollutants: A Feasible Alternative to Physical Sensors DOI Creative Commons
Martin Gabriel, Thomas Auer

Buildings, Journal Year: 2023, Volume and Issue: 13(7), P. 1684 - 1684

Published: June 30, 2023

Monitoring individual exposure to indoor air pollutants is crucial for human health and well-being. Due the high spatiotemporal variations of pollutants, ubiquitous sensing essential. However, cost maintenance associated with physical sensors make this currently infeasible. Consequently, study investigates feasibility virtually such as particulate matter, volatile organic compounds (VOCs), CO2, using a long short-term memory (LSTM) deep learning model. Several years accumulated measurement data were employed train model, which predicts pollutant concentrations based on Building Management System (BMS) (e.g., temperature, humidity, illumination, noise, motion, window state) well meteorological outdoor pollution data. A cross-validation scheme hyperparameter optimization utilized determine best model parameters evaluate its performance common evaluation metrics (R2, mean absolute error (MAE), root square (RMSE)). The results demonstrate that LSTM can effectively replace in examined room, indicating strong correlation testing set (MAE; CO2: 15.4 ppm, PM2.5: 0.3 μg/m3, VOC: 20.1 IAQI; R2; 0.47, 0.88, VOC:0.87). Additionally, transferability other rooms was tested, good CO2 mixed VOC matter 21.9 52.7 0.45, 0.09, VOC:0.13). Despite these results, they hint at potential more broadly applicable approach virtual given incorporation diverse datasets, thereby offering real-time occupant monitoring enhanced building operations.

Language: Английский

Citations

7
Calibrating low-cost sensors using MERRA-2 reconstructed PM2.5 mass concentration as a proxy DOI Creative Commons
Vasudev Malyan, Vikas Kumar, Manoranjan Sahu

et al.

Atmospheric Pollution Research, Journal Year: 2023, Volume and Issue: 15(3), P. 102027 - 102027

Published: Dec. 21, 2023

Language: Английский

Citations

4