PROJECT WORKFLOW

Our methodology for analyzing air quality across Greece follows a structured pipeline from multi-source data acquisition to WebGIS-based visualization and interpretation.

1. DATA COLLECTION

Multiple datasets were collected to support a comprehensive analysis:

  • Air Quality Monitoring Data: Monthly reanalysis data (2013–2022) for NO₂, PM₂.5, PM₁₀ from the CAMS European Air Quality Reanalysis dataset.

  • Land Cover Data:
    ESA CCI Land Cover dataset (2022), classified and
    later reclassified to IPCC classes.

    Greece Land Cover Map
  • Greece Population Classes Map

    Population Data:
    2020 population counts from WorldPop,
    used to assess exposure.

  • Administrative Boundaries:
    GAUL Level 2 (e.g., provinces) shapefile to
    support zonal and bivariate statistics.

    Greece GAUL Level 2 Map

2. DATA PROCESSING AND INTEGRATION

Preprocessing steps:

  • Clipping: All raster and vector datasets were clipped to Greece’s boundary to optimize processing.

    Greece Boundary

  • Projection: Unified CRS (WGS84 EPSG:4326) was applied to all datasets.
  • NetCDF Handling: December 2022 data was downloaded and processed manually from raw CAMS NetCDF files, then aggregated into monthly means using QGIS mesh tools.
  • Raster Aggregation:
    • Monthly to yearly average pollutant maps (2013–2022).
    • Reclassification of 2020 maps based on EU pollutant thresholds.
    • Computation of 2022 deviations from the 2017–2021 mean using Raster Calculator.

3. SPATIAL ANALYSIS AND MODELING

Key analysis steps:

  • Land Cover Reclassification: ESA CCI LC 2022 was reclassified to 11 IPCC classes using a reclassification table.
  • Zonal Statistics (Built-Up Areas): Extracted and dissolved urban zones for focused pollutant time-series stats (2013–2022).
  • Pollution Trend Analysis: Plotted NO₂/PM₂.5/PM₁₀ annual trends and extremes across land cover classes using joined zonal data.
  • Population Exposure Classes: WorldPop raster was reclassified into 5 quantile-based population classes via r.quantiles and Reclassify by table.

4. EXPOSURE AND RISK ASSESSMENT

We evaluated exposure through:

  • Pollution Class Maps: Classified 2020 pollution maps using EU thresholds.
  • Bivariate Mapping: Combined 5 pollution levels × 5 population classes into a 25-class bivariate attribute, rendered with pre-defined QML style.
  • Pie Chart Representation: Dissolved bivariate zones by pollution class, computed population sums, and visualized exposure via pie charts using DataPlotly.

5. WEBGIS IMPLEMENTATION

Interactive visualization was implemented through:

  • GeoServer Configuration: All raster and vector outputs published via WMS/WFS.
  • Layer Styling: Styled maps using .sld files including bivariate and classification symbology.
  • WebGIS & Website: Integrated using OpenLayers + HTML/CSS/JS; bivariate legends and pie charts embedded to support user interpretation.

6. VALIDATION AND QUALITY CONTROL

  • Interpolation Validation: Cross-checks between annual averages and known seasonal peaks.
  • Comparative Benchmarking: Used both EU and WHO threshold schemes for classification robustness.
  • Expert Review: Methodology aligned with lab guidance and project goals.
  • Uncertainty Consideration: Year-on-year deviation maps supported anomaly detection.