Hydrological and thermal controls of ice formation in 25 boreal stream reaches

[Display omitted] •We surveyed the ice formation in 25 boreal stream reaches.•Ice processes are very dominant in the hydrological cycle of boreal streams.•Anchor ice formed where groundwater input were low and stream power high.•We constructed a conceptual model describing the likelihood of anchor ice formation.•The study describes the complexity of ice formation in small steep boreal streams. The Northern Hemisphere has a high density of fluvial freshwater ecosystems, many of which become ice-covered during winter. The development and extent of ice have both ecological and socio-economic implications. For example, ice can cause freezing of riparian vegetation and fish eggs as well as influence hydropower production; however, when, where and why ice develops in small streams is not well known. We used observations from 25 stream reaches to study the factors controlling ice development during two consecutive winters, addressing where in the catchment surface or anchor-ice is most likely to develop, how stream morphology influences ice formation, and how climate influences ice processes. Reaches far downstream from lake outlets, or without any upstream lakes, were most prone to develop anchor-ice, but other factors also influenced ice formation. Anchor-ice was most common where water temperature and groundwater inputs were low and stream power high. Given cold air temperature and water supercooling, the in-stream substrate as well as the current velocity were also important for the development of anchor-ice. Climate and substrate seemed to be important factors for the development of surface ice. This study shows that ice processes are substantial during the hydrological year and may therefore have large implications for the ecology and engineering around boreal streams. The study also demonstrates that ice formation in the studied streams was complex, involving many variables and physical processes. We constructed a conceptual model describing the likelihood for various ice types to develop, based on the large dataset. As such, this model will be useful for practitioners and scientists working in small watercourses in the Northern Hemisphere.