Snow Depth
Last updated
Last updated
For many outdoor activities it is important to understand how much snow can be found where. For ski tourers and back country adventurers, snow is essential, while mountain bikers and hikers mostly try to avoid it as much as possible. To get a first impression, webcam feeds are used frequently. If available, they provide a good impression for parts of the trip, but are constraint by viewing angles and the topography as such. Further, actual snow depth measurements can be obtained online, for single points (at lifts in ski resorts, weather stations) or whole areas (by avalanche services, alpine clubs, meteorological service providers). In this context, it is hard to estimate the actual benefits, strengths and weaknesses of these products, since, especially in high altitude terrain, snow distribution is highly variable. Strong temperature gradients and different snow transporting processes, like wind, contribute to very heterogeneous patterns. It is therefore very time-consuming to gather all necessary information prior to a trip, estimate their meaningfulness and deduct actionable intelligence. ExoLabs took on this challenge and developed an innovative solution that is based on satellite data.
While knowledge about actual snow depths is generally more useable than the implicit snow cover, it is not yet possible to calculate it scalable based on satellite data alone. Nevertheless, a satellite based snow cover mask is essential to generate meaningful snow depth maps. Once the areal extent of snow patches is known, automatically gathered snow depth values from weather stations at various locations help to model snow depth distributions. To put single point measurements in a broader spatial context, we introduced relations to topographic features. It is easy to comprehend that very steep and sun exposed slopes will have less snow on them than flat or even bowl-shaped areas in the shade. We quantified connections of this kind with fifty datasets of the NASA "Airborne Snow Observatories".
Our interpolation technique between weather stations is based on complex geo-statistical methods that consider local and regional drifts alike. The resulting product takes into account the main influences of snow distribution: precipitation, temperature, topography, wind- and sun exposure. Its doing so by incorporating snow cover masks from satellite imagery, snow depth data from weather stations and topographic-spatial relationships.
At the moment we have high resolution snow depth data for the European Alps and Pyrenees. These maps are based on measurements from the morning and are updated daily. In case of discrepancies between the snow cover and snow depth estimations, we favour snow depth. Since fresh snow events come with clouds that block clear views for the satellites, we assume to catch these events better by prioritizing like that.
The product was validated with 23 independent reference datasets from Davos, Switzerland (2’274’247 data points in total). With an average error of just 0.36 meters snow depth, we achieved very high accuracies, considering the high variability of snow depth distributions and the spatial resolution of 20 meters.
Additionaly, we provide a global snow depth product in 300 meters resolution, based on simplified methods.
