Measurements of snowfall are very difficult and hardly satisfactory. In many countries automatic weather measurements are being made at in lieu of human observations (e.g. in the USA at most airports (1)). No automatic measurement of snow is yet replacing the manual observations carried out previously. Yet snow observations are important as regards the loads on roofs, transport problems, managing snow removal, avalanches and floods due to snow melt.
In case of snow, both the depth and the density must be measured. The best way to measure snow depth involves observations on the accumulation on a square white board of plywood. Densities of fresh snow are not measured operationally, however the layer density in mountain snow packs is often assessed for avalanche control and water supply estimates. For most applications a knowledge of the water-equivalent depth is sufficient. Heated tipping bucket gauges significantly under measure the catch, especially if no wind shields are around the gauge.
The times and frequency of measurements differ at various places, which makes comparisons of snowfalls uncertain. The measurements are very sensitive to gauge exposure, to which insufficient attention is given. In a series of tests in 1981-82, snow caught by an Alter shielded US-8'' rain gauge was studied at Barrow, at 71° N in Alaska (2). Snowfall there occurs chiefly in summer (July - October) because large parts of the Arctic ocean are ice-free. Snow was melted and the water-equivalent measured. It was determined that the yearlong loss due to evaporation of drops 'stuck' on the gauge was 12.7 mm, which was 16% of the annual total precipitation. This percentage is slightly higher in dry winter months, i.e. 15 - 22%. There were also errors due to the collection of blowing snow in the gauge, amounting to nearly half of the measured annual catch. And thirdly there was the error due to not measuring the occasional 'traces' of precipitation, equivalent to 9.3 mm in a year. In total, the US-8'' measured annual catch of 118 mm in 1982 was corrected to 189 mm, and in 1983 from 81 to 144 mm. On average, the measured catch of rain had to be enhanced by 20%, and of snow by 90%.