Explainable AI for Knowledge Acquisition in Hydrochemical Time Series V1.0.0

Abstract. The understanding of water quality and its underlying processes is important for the protection of aquatic environments. Here an explainable AI (XAI) based multivariate time series analytical framework is applied on high-frequency water quality measurements including nitrate and electrical conductivity and twelve other environmental parameters. The relationships between water quality and the environmental parameters are investigated by a cluster analysis which does not depend on prior knowledge about data structure. The cluster analysis is designed to find similar days within a cluster and dissimilar days between clusters. This allows for the data-driven choice of a distance measure. Using a swarm based AI system, the resulting cluster define three states of water bodies, which can be visualized by a topographic map of high-dimensional structures. These structures are explained by rules extracted from decision trees. The rules generated by the XAI system improve the understanding of aquatic environments. The model description presented here allows to extract meaningful, useful, and new knowledge from multivariate time series.