Operational snow mapping with simplified data assimilation using the seNorge snow model

- Revised seNorge snow model version (v.1.1.1) for operational snow mapping.
- A data assimilation procedure to correct detected model biases in near real-time.
- Pragmatic methods with modest computational and input data requirements.

SummaryFrequently updated maps of snow conditions are useful for many applications, e.g., for avalanche and flood forecasting services, hydropower energy situation analysis, as well as for the general public. Numerical snow models are often applied in snow map production for operational hydrological services. However, inaccuracies in the simulated snow maps due to model uncertainties and the lack of suitable data assimilation techniques to correct them in near-real time may often reduce the usefulness of the snow maps in operational use. In this paper the revised seNorge snow model (v.1.1.1) for snow mapping is described, and a simplified data assimilation procedure is introduced to correct detected snow model biases in near real-time. The data assimilation procedure is theoretically based on the Bayesian updating paradigm and is meant to be pragmatic with modest computational and input data requirements. Moreover, it is flexible and can utilize both point-based snow depth and satellite-based areal snow-covered area observations, which are generally the most common data-sources of snow observations. The model and analysis codes as well as the “R” statistical software are freely available. All these features should help to lower the challenges and hurdles hampering the application of data-assimilation techniques in operational hydrological modeling. The steps of the data assimilation procedure (evaluation, sensitivity analysis, optimization) and their contribution to significantly increased accuracy of the snow maps are demonstrated with a case from eastern Norway in winter 2013/2014.