Context
A study by the Council on Energy, Environment, and Water (CEEW) evaluated Delhi’s Air Quality Early Warning System (AQEWS) and Decision Support System (DSS), which forecast pollution episodes. The systems can predict high-pollution days with over 80% accuracy, but they still rely on outdated emission inventories, limiting their reliability.
Key Highlights
- High Predictive Accuracy:
- AQEWS forecasts “very poor and above” (AQI > 300) episodes with more than 80% accuracy for 2023–24 and 2024–25.
- Accuracy for “severe and above” (AQI > 400) levels stood around 70%.
- Data Outdatedness Problem:
- The system uses emission inventories from 2018, making predictions less representative of current realities.
- Delhi’s emissions profile has evolved significantly since then due to urban growth, new construction, and vehicular pattern changes.
- Forecast Model:
- The IMD and IITM jointly developed the AQEWS in 2018.
- DSS provides scenario analysis for potential pollution interventions.
- Both systems use machine learning (ML) and chemical transport models to simulate air quality outcomes.
- Study Findings:
- 83 of 92 “very poor” and “severe” AQI episodes were correctly predicted in 2023–24.
- The MAPE (Mean Absolute Percentage Error) for PM₂.₅ concentrations fell from 39% (2022–23) to 36% (2024–25) — showing improvement.
- However, reliance on old data limits the scope for short-term policy responses.
Scope for Improvement
- Update Emission Inventories:
Revise data for Delhi–NCR districts using recent satellite observations and ground-level measurements. - Model Integration:
Incorporate Graded Response Action Plan (GRAP) and seasonal meteorological variations into simulation models. - Institutional Coordination:
- The MoEFCC and CAQM should jointly fund real-time emission tracking.
- Establish data-sharing protocols across NCR states.
- Public Data Transparency:
AQEWS and DSS forecasts should be made publicly accessible and regularly updated for citizen awareness.