In context of my MSc Internship at HKVlijn in water, three snow melt algorithms are developed for Dutch snow conditions in order to include snowmelt in regional hydrology models.
Almost every winter, the Dutch lowlands are covered by a thin layer of snow. Despite its occurrence, this surface storage is currently not included in the regional water management models. Water boards are often surprised by snowfall events, and even more by its melt that results in a sudden peak discharge. Snow storage causes a delay in discharge. Current models have great difficulties to quantify this peak. Hence, the aim of this research is to develop a melt algorithm for Dutch snow conditions in order to include snowmelt in the regional hydrology models.
With this algorithm, an extra snow reservoir was implemented to the original model. Recorded precipitation was separated for rain and snow. We used snowfall as input for the reservoir, which was depleted by predicted melt. The melt was simulated using three methods: a temperature-index algorithm, a semi-empirical algorithm and an energy balance. Predicted melt and rainfall were used as input for the original hydrological model.
In this study, two models have been used (Sobek and WALRUS), which simulated discharge in the Barneveldse beek and Reusel catchment. Goodness of fit parameters were used to judge upon simulation accuracy. Moreover, a validation set with snow depth data of the KNMI verified the modelled surface reservoir.
Results show that melt algorithms improve current discharge simulations. Peak discharge is timed better, but accuracy of the absolute drained discharge differs per winter. In general, best performances are achieved using the two (semi-) empirical algorithms.
The energy balance algorithm seems to be sensitivity to rain-on-snow events. Hence, it is crucial to separate precipitation correctly. In addition, spatial variation plays an important role for accurate simulation of the energy balance fluxes (e.g. albedo, cloud cover).