A geospatial model to determine patterns in river ice cover breakup and jamming behaviour

Abstract

In the past, both empirical and process-based attempts have been made to predict river ice behaviour, in particular ice cover breakup and ice jamming occurrences. These methods perform with varying and limited success and tend to be site specific. A method is required which can simply estimate the predisposition of river reaches to ice breakup and jamming events. This paper introduces a geospatial modelling approach which can fulfil that task and improve the predictive power of ice cover breakup and ice jamming behaviour. The geospatial model can determine the most vulnerable sections along the studied reaches to such behaviour, which are phenomena entailing hydraulic, ice morphology and fluvial geomorphology. A geospatial model clusters hydraulic characteristics (e.g. discharge or stage), ice characteristics (e.g. ice thickness and ice type) and river geomorphological characteristics (e.g. sinuosity, slope, width, etc.) into common river features called Geomorphic Response Units (GRU). A statistical clustering technique such as principle component analysis (PCA) is used to derive these GRUs. It is assumed that certain GRUs will be more susceptible to certain ice cover behaviour, such as breakup and jamming of river ice. Data acquired along the Slave River and its delta in Canada is used to test the geospatial model. The main data sources are space-borne remote sensing MODIS imagery and traditional and local knowledge from members of the communities alongside the river, in particular Fort Resolution and Fort Smith