A New Data-Driven Framework for Operationalising Complexity in Contaminated-Site Management

Contaminated site management programmes often rely on risk classification and pollutant metrics, yet these tools provide limited insight into the practical difficulty of characterising, remediating and closing sites. This paper introduces a modular and data-driven framework that operationalises contaminated-site complexity as a distinct managerial dimension, and implements it through a Contaminated Site Complexity Index (CSCI) that can be computed from routine registry fields, including in settings where data are heterogeneous or incomplete.The methodological contribution of the paper lies in the formalisation of the CSCI framework and its deployment on real-world registry data from the State of São Paulo, Brazil. Using a clustering-based approach, the framework identifies three empirically derived complexity tiers within the registry, enabling a structured interpretation of how complexity is distributed across contaminated sites. Validation against a subset of institutionally recognised critical sites shows that these sites systematically fall within the high and upper-medium complexity tiers, indicating alignment between the CSCI outputs and expert judgement embedded in regulatory practice. A longitudinal analysis further demonstrates that extended remediation timeframes are predominantly associated with medium- and high-complexity sites, providing additional empirical support for the proposed complexity classification.The results show that the CSCI framework generates policy-relevant insights even when applied to non-exhaustive datasets, highlighting its robustness and practical utility. The framework can be used either as a standalone analytical tool or as a complementary layer alongside risk-based assessments. More broadly, the CSCI offers a transferable governance instrument applicable across jurisdictions with registries at different stages of maturity, while also providing a conceptual scaffold for countries where contaminated-site databases are still absent or under development.