How well do modern snow-mapping models stack up against measurements from people like you? Researchers have taken a look at 3 products, each published daily: the Interactive Multisensor Snow and Ice Mapping System (IMS) from NOAA, the Canadian Meteorological Centre’s (CMC) Daily Snow Depth Analysis Data, and the Snow Data Assimilation System (SNODAS) from the NOAA National Weather Service. IMS shows snow and ice cover, but not depth, for the Northern Hemisphere. CMC and SNODAS provide snow depth data on a daily basis, with CMC covering the Northern Hemisphere at a moderate resolution (24 x 24 km grid cell) and SNODAS covering the continental United States at a higher resolution (1 x 1 km grid cell). The results are shown opposite. Figure 1 shows the number of snow depths greater than 0 cm reported by Snowtweets and how they match the IMS snow cover class. In addition, it is important for observers to submit 0 snow depth measurements (or snow free land) which also agree with the IMS snow free category. Other less obvious reports are located over water where the IMS snowmap cannot resolve land on a small island in a water body or where a reported snow depth was over ice.
Snow measurements from people across the world have been instrumental in validating these simulation models. From the simple comparisons made to date, the Snowtweets and the models are telling the same story! At the higher resolution grid, SNODAS agrees better with the very high resolution Snowtweets data (Figure 2) than the CMC data (Figure 3). This tells us something about the nature of snow depth variability and how we should be monitoring it – frequently and at high resolution! With more tweets in future winters, real-time snow mapping can be performed, creating an instrumental tool in weather and climate research.
Figure 1. The IMS product is a daily snow and ice cover map over the Northern Hemisphere. Comparing Snowtweets measurements to the predicted cover as a presence/absence test, snow measurements are consistent with snow cover estimates. The outliers represent snow fall events, variations in local topography or areas which border coastal regions.
Figure 2. The NOAA National Weather Service’s 1 x 1 km SNODAS product estimates of snow depth across the United States based on satellite observations, ground stations and numerical simulation models. On average, snow measurements from people around the United States are consistent with SNODAS estimates, with differences due to local variation and snowfall events. The narrower spread of data in both x and y axes demonstrates that the Snowtweets and the model are agreeing well.
Figure 3. The Canadian Meteorological Centre (CMC) Daily Snow Depth Analysis Data product estimates snow across the Northern Hemisphere based on measurements made at human operated and automatic weather stations in the WMO global weather station network. The wide spread relfects the coarse grid cell sizes used in the model relative to the local scale measurements of Snowtweets.