With the increasing use of numerical models predicting the snow cover status, the need for a simple and standardized evaluation procedure arises. We present such a method that compares numerical model profiles with snow pit profiles and provides a quantitative statistical agreement-disagreement measure. The method can also be used to compare several model profiles with each other.
The first step of the method is the mapping of the model profile layers onto the layers of the observed profile. This mapping is necessary to adjust for deviating total snow depth and shifted positions of the layers. Following the mapping, the individual profile parameters such as grain type, grain size, liquid water content, temperature and density are compared. The result of the comparison is a score between 0 (profiles show no agreement) and 1 (profiles are identical) for each parameter. The parameter scores can be combined to give an overall profile score between 0 and 1. The method facilitates evaluation studies and allows to quantify improvements made in the modeling of processes in the snow cover.
This is illustrated by analyzing a simulated profile at an IMIS (German: Interkantonales Mess-und Informations System) automatic snow and weather station with an observed, detailed snow pit profiles from the Swiss Alps. The agreement score for the parameter grain type increased from 0.3 to 0.7 after introducing an additional grain type in the simulation. In addition, the scores were calculated for two observation stations in the Swiss Alps for a whole winter season and it was detected that for most of the winter an overall agreement score between SNOWPACK simulation and snow pit profile of approximately 0.8 is reached. The temperature regime is modeled best and most difficulties are encountered with grain type. An important result is further that the energy balance processes at the beginning of the spring melt season has to be improved.