Modelling river flow in cold and ungauged regions: a review of the purposes, methods, and challenges

River flow forecasting models assist in the understanding, predicting, monitoring, and managing of issues related to surface-water resources, such as water quality deterioration flooding, or developing adaptation strategies cope with climate change increasing demand. This review presents an overview current research status progress river-flow forecasting, focusing on cold climates ungauged locations. River-flow regions represents a challenge because natural processes that occur within catchments vary greatly both seasonally annually. variability, which highly depends climatic topo-geomorphological characteristics basin, translates into increased model uncertainty substantial limitation when attempting forecast river regions, are often poorly gauged ungauged. To address this limitation, “Predictions Ungauged Basins” initiative offers variety studies improve performance by adopting regionalization, spatial calibration, interpolation, regression approaches. Process-based demonstrate significant improvement including remote-sensing data replicate derive complex hydrological processes. Empirical models, utilize observed formulate graphical solution, unlike mathematical require formulating relationships between processes, also implemented most recent developments machine learning, showing exceptional accuracy. Although process-based provide wide understanding watershed hydrology, unavailable, expensive, time-consuming collect. They generate numerous calibration parameters, resulting computationally demanding methods operate. using empirical reduces number parameters but could produce biased results insufficient variables available explain physical mechanisms watershed’s hydrology. Moreover, be potentially sensitive validation dataset selection. In review, Canadian primarily selected highlight some efforts may necessary other similar including: (i) coping limited availability through regionalization methods; (ii) providing user-friendly interfaces; (iii) advancing structure; (iv) universal method for transferring parameters; (v) standardizing selection; (vi) integrating models.