Sentinel-5P TROPOMI SO2 Analysis

End-to-end atmospheric monitoring of sulfur dioxide over the Philippines, combining satellite remote sensing with ground validation and public health integration.

--
QA-filtered pixels
--
Days of coverage
24
Volcanoes monitored
--
Eruption events

Data Pipeline

Automated ingestion, quality-controlled processing, and multi-stage scientific analysis of Sentinel-5P Level 2 SO2 column density retrievals.

NASA GES DISC S5P_L2__SO2____ v02.02
download_so2.sh 526 granules, 455 MB
process_so2.py QA > 0.5 filter
Parquet 168 MB, ~12M pixels
8-Stage Analysis Quality → Catalogue
Technical Notes
DOAS Retrieval
The TROPOMI SO2 product uses Differential Optical Absorption Spectroscopy (DOAS) in the UV (312–326 nm). The 1 km box air mass factor (AMF) assumes SO2 within the lowest 1 km of the atmosphere — appropriate for anthropogenic and passively degassing volcanic sources, but may underestimate column density for elevated eruption plumes.
QA Filtering
Only pixels with qa_value > 0.5 are retained, removing cloud-contaminated and low-quality retrievals. This threshold balances data coverage against noise. Notably, valid volcanic signals can be lost during heavy ash loading, as the DOAS fit quality degrades.
Negative Pixels
Negative SO2 retrievals (~40% of clear-sky pixels) are retained in all statistics. These represent instrument noise around zero background and are essential for unbiased mean estimates.

Data Quality & Coverage

Daily temporal coverage and SO2 column density over the Philippines for 2025. The dual-axis time series reveals data gaps from orbit geometry and the inverse relationship between coverage and mean SO2.

Daily Pixel Count & Mean SO2
Loading chart…

Spatial Climatology

Annual mean SO2 column density on a 0.1° grid, showing persistent enhancement near active volcanoes and the low background across open ocean areas.

Gridded Annual Mean SO2 (Top 1000 cells by mean DU)
Loading map…

Volcano Monitoring

Systematic monitoring of 24 Philippine volcanoes using TROPOMI SO2 column density within 50 km radii. Ranking by maximum observed SO2 highlights the most active sources.

Volcano Activity Map
Loading map…
Maximum SO2 Ranking
Loading chart…
Volcano Time Series (7-day rolling mean)
Loading chart…

Eruption Forensics: Kanlaon Case Study

Kanlaon volcano (Negros Island) produced sustained SO2 emissions throughout 2025. Event-level analysis captures plume spatial extent, mass loading, and decay behavior.

Event Timeline
Loading chart…
Plume Spatial Extent
Loading map…
SO2 Decay Curve
Loading chart…
Event Mass Loading Comparison
Loading chart…

Urban & Anthropogenic SO2

Regional SO2 enhancement above ocean background for Metro Manila, Cebu Metro, Batangas, Iligan City, and Cagayan de Oro. Detection signal-to-noise ratio (SNR) indicates statistical significance of excess SO2.

Region vs. Ocean Background
Loading chart…
Detection Signal-to-Noise Ratio
Loading chart…
Monthly Excess SO2
Loading chart…

Public Health Integration

Preliminary framework for integrating satellite SO2 observations with population exposure assessment, WHO guideline thresholds, and ground-based air quality monitoring.

WHO Guideline Thresholds

Important caveat TROPOMI measures total column SO2 (in Dobson Units), not surface-level concentration. Converting column density to ground-level concentration (µg/m³) requires atmospheric modeling (e.g., CAMS, GEOS-Chem) accounting for boundary layer height, vertical SO2 profile, and meteorology. The thresholds shown here are approximate, assuming uniform mixing within a 500–1000 m boundary layer.
Daily SO2 with WHO Threshold Uncertainty Band (0.7–1.4 DU)
Loading chart…

Population Exposure

Regional Population-Weighted SO2 Exposure
Loading chart…
Population in SO2 Monitoring Regions (2020 Census)
Region Population Mean SO2 (DU) Annual DU-days

DU-days = mean column SO2 (DU) × 365 days. This is a satellite-derived exposure proxy, not a health risk metric.

Ground Validation

OpenAQ Monitoring Stations vs. TROPOMI Coverage
Loading map…
Validation challenges No SO2 ground monitoring stations were identified in the Philippines via OpenAQ. PM2.5 stations are shown instead to illustrate the monitoring network. Satellite-ground comparison for SO2 is limited by: overpass timing (~13:30 local), column vs. surface measurement, and spatial resolution mismatch (5.5×3.5 km pixel vs. point station).

Methodology & Limitations

Transparency in analytical choices and known limitations is essential for scientific credibility. The following issues are documented and tracked.

Critical Open Issues
Issue Impact Mitigation
SO2 column product (1 km AMF) Underestimates elevated volcanic plumes Consider 7/15 km AMF variants for eruptions
Enhancement threshold (1 DU) vs. noise floor TROPOMI noise ~0.5–1.5 DU; potential false positives Statistical detection via SNR-averaged metric
Fixed pixel area (19.25 km²) Actual 19–56 km²; up to 4× mass loading error at swath edges Report mass loading as lower bound
QA > 0.5 for volcanic events Dense aerosol may invalidate otherwise valid pixels Evaluate lower QA thresholds during eruptions
Ocean background representativeness Single 2×2° box; potential ship/volcanic contamination Multi-region background or latitudinal stratification
Analysis Parameters
Parameter Value Rationale
QA threshold > 0.5 Standard TROPOMI L2 recommendation
Enhancement threshold 1.0 DU Above noise floor for aggregated statistics
Strong volcanic threshold 5.0 DU Clear volcanic signal above all noise sources
Volcano monitoring radius 50 km Captures near-source plumes within ~1 TROPOMI swath width
Minimum reliable pixels 5 Statistical minimum for daily mean estimate
DU conversion 4.4615 × 10-4 mol/m² Standard Dobson Unit definition