Machine Learning-Based Approach Using Open Data to Estimate PM2.5 over Europe

Ibrahim, Saleem; Landa, Martin; Pešek, Ondřej; Brodsky, Lukas; Halounová, Lena

doi:10.3390/rs14143392

Cited by 17 publications

(9 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this study, we used a tree-based machine learning algorithm called the Extra Trees to estimate PM2.5 over CZ with a high spatial resolution of 1 km during 2018 ̶ 2020. In contrast to our prior study, which concentrated on the entire Europe [31], we discovered that incorporating more balanced data in terms of spatial and temporal distribution enhances the overall accuracy of the model and simplifies the modeling approach. The R2 obtained from the 10-fold cross-validation of the model developed specifically for CZ was 0.85, whereas the corresponding R2 for the model developed for the entire European region was 0.69 [31].…”

Section: Discussioncontrasting

confidence: 86%

“…ET reduces overfitting by introducing additional randomness during the construction of the trees and it uses the entire dataset while training without performing any pruning which decreases the required time for training compared to the RF that applies pruning techniques. A deeper explanation of this algorithm was provided in our previous work [25,31].…”

Section: Model Developmentmentioning

confidence: 99%

See 1 more Smart Citation

PM2.5 Estimation in the Czech Republic using Extremely Randomized Trees: A Comprehensive Data Analysis

Ibrahim,

Landa,

Matoušková

et al. 2023

CEJ

View full text Add to dashboard Cite

The accuracy of artificial intelligence techniques in estimating air quality is contingent upon a multitude of influencing factors. Unlike our previous study that examined PM2.5 over whole Europe using unbalanced spatial-temporal data, the focus of this study was on estimating PM2.5 specifically over the Czech Republic using more balanced dataset to train and evaluate the model. Moreover, the spatial autocorrelation between the ground-based station was taken into consideration while building the model. The feature importance while developing the Extra Trees model revealed that spatial autocorrelation had greater significance in comparison to commonly used inputs such as elevation and NDVI. We found that R2 of the 10-CV for the new model was 16% higher than the previous one. R2 reached 0.85 when predicting unseen data in new locations. The developed spatiotemporal model was employed to generate comprehensive daily maps covering the entire study area throughout the 2018–2020 years. The temporal analysis showed that the levels of PM2.5 exceeded recommended limits of 20 µg/m3 during the year 2018 in many regions. The eastern part of the country suffered from the highest concentrations especially over Zlín and Moravian-Silesian Regions where in the 2018 winter, the values reached risky average concentrations of 30 µg/m3 and 35 µg/m3 respectively. Air quality improved during the next two years in all regions reaching promising levels in 2020 where almost all regions had average concentrations less than 20 µg/m3. The generated dataset will be available for other future air quality studies.

show abstract

Section: Discussioncontrasting

confidence: 86%

Section: Model Developmentmentioning

confidence: 99%

PM2.5 Estimation in the Czech Republic using Extremely Randomized Trees: A Comprehensive Data Analysis

Ibrahim,

Landa,

Matoušková

et al. 2023

CEJ

View full text Add to dashboard Cite

show abstract

“…Spatial open datasets differ in quality and quantity and are used for various types of scientific and practical research in order to be exploited for various purposes, such as, for example, online applications [45], 3D city modelling [46], for collaborative geological mapping [47], for investigating historical settlements and landscape analysis [48], to study the geographic educational paths of individuals or social groups [49], for building citizen science [50], analysis of air pollution [51], vehicular traffic [52], public green spaces [53], etc. Special attention is given to the comparative analysis of open data among cities in compliance with indicator standards in order to set up a common set of indicators [54], such as economy, education, energy, environment and climate change, finance, governance, health, housing, population and social conditions, recreation, safety, solid waste, sport and culture, telecommunications, transportation, urban/local agriculture and food security, urban planning, wastewater and water.…”

Section: Discussionmentioning

confidence: 99%

Open Data as a Condition for Smart Application Development: Assessing Access to Hospitals in Croatian Cities

et al. 2022

View full text Add to dashboard Cite

This research aims to assess available spatial open data related to access to hospitals in the three largest Croatian cities (Zagreb, Split, Rijeka), with a future aim to create digital services as an ecosystem that will be used in everyday situations, as part of the concept of “digital society”. Data analysis is performed for the following datasets: hospitals, hospital specialization, public transport (tram, bus), bike routes, car routes, parking and parking for people with disabilities. The future aim is to create a new mobile, multilingual and voice-based application that would enable quick access to information on hospital access, relying on the principles of the open data ecosystem, which would improve over time. There are four specific aims: (1) to identify and analyze portals and open datasets of the selected categories for the three largest Croatian cities; (2) to analyze existing open data assessment frameworks and detect gaps; (3) to create a conceptual open data assessment framework as an open data ecosystem that integrates new end-user perspectives; and (4) to analyze existing open datasets for the three largest cities in Croatia, based on supply and demand by researcher/developer categories, using the newly developed framework. The results show that existing open datasets related to hospital access in the three largest Croatian cities are scattered across different portals. Analyzed frameworks existing on the supply side of open data are more focused on the status of the components, lacking evaluation scales and not including end-user-driven aspects, which would be crucial for the open data ecosystem. As a result, the new “Hospital Access Framework” is created as a conceptual ecosystem, including five categories: supply, demand by researcher/developer, demand by end-user, legal aspect and impact. Analysis of existing open data for the three Croatian cities is performed for two categories (supply and demand by researcher/developer), for which KPIs, indicators and evaluation scales are developed. The other three categories are not analyzed, since the application cannot be developed from existing data, which are insufficient for the creation of a smart application. Results show that existing open data related to hospital access are incomplete or do not exist at all (hospital specialization, parking for people with disabilities, data on multilingualism and voice enabling). Average scores of the supply category are higher than those for demand by researcher/developer, although they are below the average grade, showing a lack of available data that could be used for further development. The new conceptual “Hospital Access Framework” open data ecosystem would benefit from end-user interaction, therefore, improving over time with end-users through interaction.

show abstract

“…10). Examples are the potential and realized distribution of 16 tree species (Bonannella et al, 2022), monthly airborne fine particulate matter levels (Ibrahim et al, 2022), 43 CORINE land cover classes (Witjes et al, 2022), and daily aerosol optical depth levels (Ibrahim et al, 2021).…”

Section: Future Workmentioning

confidence: 99%

Ecodatacube.eu: Analysis-ready open environmental data cube for Europe

Witjes¹,

Parente²,

Križan

et al. 2023

Preprint

View full text Add to dashboard Cite

The paper describes the production steps and accuracy assessment of an analysis-ready, open-access European data cube consisting of 2000–2020+ Landsat data, 2017–2021+ Sentinel-2 data and a 30m resolution Digital Terrain Model (DTM). The main purpose of the data cube is to make annual continental-scale spatiotemporal machine learning tasks accessible to a wider user base by providing a spatially and temporally consistent multidimensional feature space. This has required systematic spatiotemporal harmonization, efficient compression, and imputation of missing values. Sentinel-2 and Landsat reflectance values were aggregated into four quarterly averages approximating the four seasons common in Europe (winter, spring, summer and autumn), as well as the 25th and 75th percentile, in order to retain intra-seasonal variance. Remaining missing data in the Landsat time-series was imputed with a temporal moving window median (TMWM) approach. An accuracy assessment shows TMWM performs relatively better in Southern Europe and lower in mountainous regions such as the Scandinavian Mountains, the Alps, and the Pyrenees. We quantify the usability of the different component data sets for spatiotemporal machine learning tasks with a series of land cover classification experiments, which show that models utilizing the full feature space (30 m DTM, 30 m Landsat, 30 m and 10 m Sentinel-2) yield the highest land cover classification accuracy, with different data sets improving the results for different land cover classes. The data sets presented in the paper are part of the EcoDataCube platform, which also hosts open vegetation, soil, and land use / land cover (LULC) maps created. All data sets are available under CC-BY license as Cloud-Optimized GeoTIFFs (ca. 12 TB in size) through SpatioTemporal Asset Catalog (STAC) and the EcoDataCube data portal.

show abstract

Machine Learning-Based Approach Using Open Data to Estimate PM2.5 over Europe

Cited by 17 publications

References 45 publications

PM2.5 Estimation in the Czech Republic using Extremely Randomized Trees: A Comprehensive Data Analysis

PM2.5 Estimation in the Czech Republic using Extremely Randomized Trees: A Comprehensive Data Analysis

Open Data as a Condition for Smart Application Development: Assessing Access to Hospitals in Croatian Cities

Ecodatacube.eu: Analysis-ready open environmental data cube for Europe

Contact Info

Product

Resources

About