HYDROLOGIC ASSESSMENT OF A SMALL MARITIME HYDROMETRIC MONITORING NETWORK USING A PROCESS-GUIDED CATCHMENT CLASSIFICATION SYSTEM

Abstract

The National Hydrometric Program, operated by the Water Survey of Canada, is the primary source of surface water quantity data in Canada. The network is cost-shared between the federal and provincial governments, and decisions relating to station placement are made according to both federal and provincial interests. Nova Scotia is a small maritime province in Atlantic Canada with a relatively diverse climate and geology for its small size. The Nova Scotia hydrometric monitoring network currently consists of 31 stations. The overall objective of this study was to determine how well the current network captures the level of hydrologic variability expected in the province using a regional catchment classification scheme. To accomplish this, we developed a combined inductive-deductive catchment classification system and applied it to the province’s active monitoring network and 246 ungauged major watersheds. Initially, hydrologic signatures were used to quantify the catchment function of 47 long-term gauged catchments and to cluster similarly behaving catchments. We identified five generalized flow classes and then attempted to replicate this classification using a deductive-based decision tree framework with physiographic and meteorological explanatory variables. The validated decision tree was used to classify the active hydrometric network and 246 major watersheds in the province. The network was assessed to determine how well it covered the expected hydrologic variability in the major watersheds across the province. The results indicated that current active hydrometric monitoring network does not adequately capture the range of hydrologic variability within this region; three of the five primary flow regimes observed in the province are not well represented by the active viii network. The analysis also illustrated that the stations in the active monitoring network are not evenly distributed among the five generalized flow classes seen in Nova Scotia. The decision tree proved to be a useful tool for understanding the current network’s coverage and could also be easily applied by practitioners to identify appropriate donor catchments for ungauged watersheds.