Conventional Modelling versus Knowledge-based Prediction for Pollution Remediation

Increasing water scarcity in many developing countries is forcing investments into remediation of water quality at the basin or sub-basin scale in order to increase water availability. Remediation decisions involving complex aquatic environments are often made in data-poor and knowledge-poor situations. Remediation objectives are often poorly articulated, raise unrealistic expectations, and cannot be evaluated in cost-benefit terms. Mathematical modelling, as a means of determining remediation options, is the usual method of choice in data-rich developed countries and requires substantial investment in reliable data, scientific capacity and a sophisticated management culture that generally are not found in developing countries. Modelling is expensive, has numerous other technical problems in developing countries, requires a high degree of input by foreign experts, and rarely leaves residual capacity in the developing country. In contrast, new techniques in knowledge-based (K-B) prediction focus on use of local and domain knowledge to establish meaningful program objectives. K-B-based decision support systems (DSS) allow the client to game with alternative remediation options with outputs expressed in degree of uncertainty in the assumptions and analytical processes included in the DSS system. The K-B approach builds local capacity and, by providing access to domain knowledge, reliance on foreign experts diminishes as local experts assume similar tasks elsewhere in the country.