Retrieval of river ice parameters for ice jam prediction with MODIS on the lower Susquehanna
During wintertime (and sometimes, midwinter), rivers may transition from fully open water to partially or fully frozen. Ice jams are caused by sometimes sudden ice accumulations in rivers, especially when their movement is arrested due to an impeding ice sheet or other obstructions. As result, they likely may restrict the flow cross section and cause flooding due to backwater effects. Furthermore, these jams may break suddenly, resulting in a significant surge wave that impacts both up- and downstream locations. Mechanical break-up is difficult to predict, even with current models, and in-situ measurements are difficult. Generally, ice-jams have been examined with particular focus on hydrometric stations only, but recent studies indicate that valuable information may be gleaned by incorporating data from larger regions in the analysis, such as data obtained at up- or downstream hydrometric stations. In order to better retrieve ice parameters over a larger region, this work employs the Moderate Resolution Imaging Spectroradiometer (MODIS) to assess where, when and how much ice is present in the river stretch. As a first attempt, the Crest River Ice Observation System (CRIOS) was used in combination with a GLM model to predict ice jam events. It was found that the MODIS cloud mask causes problems for the CRIOS river ice detection algorithm. Hence, this work presents an alternative approach that is expected to result in improved river ice detection and classification – and will form the basis of a jam prediction model. The study area consists of a 50 km stretch on the Susquehanna River, and is centered Harrisburg, PA and looks at the 2013-2014 winter season. Satellite images can corroborate in-situ observation of freeze-up and break-up of ice and river-ice for the 2014 winter. Furthermore, satellite images show that river ice can be detected from the beginning of January to the middle of March 2014.