It is estimated that seasonal snow cover is the primary source of water supply for over 60 million people in the western United States. Informed decision making, which ensures reliable and equitable distribution of this limited water resource, thus needs to be motivated by an understanding of the physical snowmelt process. We present a direct application of hybrid systems for the modeling of the seasonal snowmelt cycle, and show that through the hybrid systems framework it is possible to significantly reduce the complexity offered by conventional PDE modeling methods. Our approach shows how currently existing heuristics can be embedded into a coherent mathematical framework to allow for powerful analytical techniques while preserving physical intuition about the problem. Snowmelt is modeled as a three state hybrid automaton, representing the sub-freezing, sub-saturated, and fully saturated physical states that an actual snowpack experiences. We show that the model accurately reproduces melt patterns, by simulating over actual data sets collected in the Sierra Nevada mountains. We further explore the possibility of merging this model with a currently existing wireless sensing infrastructure to create reliable prediction techniques that will feed into large scale control schemes of dams in mountain areas.
Kerkez B., Glaser S., Dracup J., Bales R. (2010): A Hybrid System Model of Seasonal Snowpack Water Balance. HSCC '10 Proceedings of the 13th ACM international conference on Hybrid systems: computation and control, Stockholm, Sweden.