Error identification/propagation/remediation in biomonitoring surveys—A knowledge-based approach towards standardization via fault tree analysis

Biomonitoring has become a key concept in environmental management since it is the most ecologically-relevant means for assessing pollution impact. Its broad applicability, however, raises the need for harmonization, optimization and standardization. The main difficulty met in the development of a generalized methodological framework for standardizing biomonitoring surveys is to reconcile a theoretical approach with an operational approach: in any set-up the survey strategy should ensure that the measured values represent the status of the environment. This leads, inevitably, to the application of a variety of methods, techniques and strategies in order to accommodate the special ecogeomorphological characteristics of each area and handle adequately the knowledge gaps related to local species stress response mechanisms and tolerances. Thereby, comparability of the results, even between subsequent surveys in the same area or concurrent surveys at neighbouring areas, is unfeasible, yet indispensable in defining spatiotemporal pollution patterns in large ecosystems. This inevitably requires some kind of normalization/harmonization that would strengthen any observed correlations between exposure and health effects, which ultimately may point at potential causal relationships. The aim of this work is to design/develop a knowledge management tool, built on a cybernetic infrastructure for (i) localizing the variation source(s) in each project that prohibit inter-survey normalisation/comparability, (ii) determining the path of error propagation as a causal chain when a fault is identified, (iii) testing the ultimate causes suggested as mostly responsible for this fault, and (iv) proceeding to remedial proposals (including a feedback possibility in case that the suggested remedy is proved to be inadequate) with a view to improving quality and reliability of biosurveillance. The presented tool relies on Fuzzy Fault Tree Analysis (FFTA) to identify, categorise, sort and analyse all possible sources of variation and error in biomonitoring; thereby, an expert system is developed, where the tree (dendritic) structure serves as the Knowledge Base (KB) and the fuzzy rules based decision mechanism is the inference engine. This scheme, relying on a collaborative model building methodology and a systemic modeling formalism by using 2nd order cybernetics in order to include human judgement and reasoning, enables knowledge to be used not only for representation but also for re