Spatially constrained clustering of ecological units to facilitate the design of integrated water monitoring networks in the St. Lawrence Basin. Issue 2 (1st February 2016)

• Record Type: Journal Article
• Authors: Adams, Matthew D.; Kanaroglou, Pavlos S.; Coulibaly, Paulin
• Is Part Of: International journal of geographical information science. Volume 30:Issue 2(2016)
• Page Start: 390
• Page End: 404
• Publication Date: 2016-02-01
• Subjects: Ecological units -- regionalization -- water monitoring network -- principal component analysis -- spatial analysis; Geography -- Data processing -- Periodicals; Information storage and retrieval systems -- Periodicals; Géomatique -- Périodiques; Systèmes d'information -- Périodiques; 910.285
• DOI: 10.1080/13658816.2015.1089442
• Languages: English
• ISSNs: 1365-8816

Abstract

ABSTRACT: Water monitoring networks are generally classified into surface water, precipitation, groundwater or water quality monitoring networks. The design of these networks typically occurs in isolation from each other. We present a regionalization approach to identify homogeneous subregions of large basins that are suitable as areas for the optimization of an integrated water monitoring network. The study area, which comprises a portion of the St. Lawrence Basin, was spatially divided using ecological units. For each ecological unit, 21 attributes were derived including both environmental and hydrological indicators. A spatially constrained regionalization technique was applied to define the final regions. A scree plot was used to determine the number of regions. The sensitivity of the technique to the correlation in the attribute data was removed by utilizing principal component analysis to reduce correlation between attribute data. During regionalization, the component values were weighted by their proportion of the total variance explained. The four regions in the final configuration had areas from 19% to 31% of the total area, 63, 597 km 2 . For the St. Lawrence Basin, this approach is effective for defining homogeneous regions that can be used in further research on the optimization of integrated water monitoring networks. The approach is portable to other regions and can incorporate any set of attribute data that is valuable to the regionalization objective.