Weather as the Conductor of Pollution-Treatment Plants

A three-phase progressive protocol (in silico → minimal → full) is used to validate the reproducibility of regimes and to reduce the risk of costly late-stage failure. This sequential approach is considered a model of good practice in experimental design.

The hypothesis proposes an elegant and operationalisable transition from diagnostic climatological classification (clustering) to a proactive real-time control strategy, which constitutes a significant conceptual advance relative to the state of the art, which is often limited to pattern analysis.

The idea of using hierarchical clustering to capture recurrent weather regimes is conceptually interesting and builds upon existing literature in climate classification.

Immediate captive market: Metropolitan areas that do not comply with WHO PM2.5 standards (e.g., Delhi NCR, Beijing-Tianjin-Hebei, Silesia in Poland) have regional authorities (e.g., CPCB in India, MEE in China) prepared to pay for dynamic control software solutions that avoid costly plant shutdowns. The addressable market is estimated at 50–100 M€ per year for software licences alone.

Strong conceptual originality: coupling unsupervised meteorological clustering with distributed pollution control is a little-explored approach, with potential for a step-change advance in smart cities.