Programas de monitoramento emergencial de qualidade da água : redesign de rede através de métodos baseados na Entropia de Shannon

Abstract

The occurrence of large-scale environmental disasters has placed environmental management, particularly water resource management, at the forefront of global attention. Depending on the magnitude of the impacts and their geographical extent, the process of environmental recovery becomes quite complex and challenging, often requiring the development of solutions that are not readily available. In the case of water resource management, the entire chain of actions necessary for the recovery of impacted ecosystems involves establishing programs and networks for monitoring water quality. Despite its importance, emergency monitoring networks have been underexplored in studies focused on resizing monitoring networks. Even less common has been the use of approaches based on Shannon’s Entropy, although there is significant potential for its application. This research evaluated the suitability of methods for resizing monitoring networks based on Shannon’s Entropy, aiming to support the enhancement of emergency water quality monitoring programs in watersheds affected by environmental disasters. The study case focused on the Doce River watershed, which was impacted by the rupture of a mining waste dam. Methods based on the analysis of the information content of water quality variables monitored by an emergency network were applied to select the set of variables referred to as "non-critical." Furthermore, through the application of an original optimization methodology, an effort was made to select the set of stations corresponding to the arrangement of the best performing network. The prior segmentation of the watershed, through cluster analysis, allowed for the incorporation of underlying socio-environmental characteristics of the sub-regions into the analyses, in addition to reducing the complexity of the optimization problems. However, it imposed the necessity to clearly demonstrate the effectiveness of the method and the impact of segmenting the networks. "Non-critical variables" were selected in proportions ranging from 32% to 50% in each sub-region of the Doce River watershed, illustrating opportunities to reduce the frequency of monitoring these variables. The implemented methodology was able to capture the spatiotemporal variability of the informational content of the variables, even those that may be present in very small proportions in the environment. Regarding the spatial configuration of the analyzed network, there is potential for a 20% reduction of the stations, without compromising the generation of useful information for water quality management. However, the results suggest that the proposed optimization method requires improvement to enhance its performance in networks with a significant number of stations and to make it more objective. Decision-makers may find in these methods strategies to support the management of emergency water quality monitoring programs, potentially improving network performance and the cost-benefit ratio of data generation.