Improving snowfall forecasting by diagnosing snow density

Abstract

Current prediction of snowfall amounts is accomplished either by using empirical techniques or by using a standard modification of liquid equivalent precipitation such as the 10-to-1 rule. This rule, which supposes that the depth of the snowfall is 10 times the liquid equivalent (a snow ratio of 10:1, reflecting an assumed snow density of 100 kg m-3), is a particularly popular technique with operational forecasters, although it dates from a limited nineteenth-century study. Unfortunately, measurements of freshly fallen snow indicate that the snow ratio can vary from on the order of 3:1 to (occasionally) 100:1. Improving quantitative snowfall forecasts requires, in addition to solutions to the significant challenge of forecasting liquid precipitation amounts, a more robust method for forecasting the density of snow. A review of the microphysical literature reveals that many factors may contribute to snow density, including in-cloud (crystal habit and size, the degree of riming and aggregation of the snowflake), subcloud (melting and sublimation), and surface processes (compaction and snowpack metamorphism). Despite this complexity, the paper explores the sufficiency of surface and radiosonde data for the classification of snowfall density. A principal component analysis isolates seven factors that influence the snow ratio: solar radiation (month), low- to midlevel temperature, mid- to upper-level temperature, low- to midlevel relative humidity, midlevel relative humidity, upper-level relative humidity, and external compaction (surface wind speed and liquid equivalent). A 10-member ensemble of artificial neural networks is employed to explore the capability of determining snow ratio in one of three classes: heavy (1:1 < ratio < 9:1), average (9:1 ≤ ratio ≤ 15:1), and light (ratio > 15:1). The ensemble correctly diagnoses 60.4% of the cases, which is a substantial improvement over the 41.7% correct using the sample climatology, 45.0% correct using the 10-to-1 ratio, and 51.7% correct using the National Weather Service "new snowfall to estimated meltwater conversion" table. A key skill measure, the Heidke skill score, attains values of 0.34-0.42 using the ensemble technique, with increases of 75%-183% over the next most skillful approach. The critical success index shows that the ensemble technique provides the best information for all three snow-ratio classes. The most critical inputs to the ensemble are related to the month, temperature, and external compaction. Withholding relative humidity information from the neural networks leads to a loss of performance of at least 5% in percent correct, suggesting that these inputs are useful, if nonessential. Examples of pairs of cases highlight the influence that these factors have in determining snow ratio. Given the improvement over presently used techniques for diagnosing snow ratio, this study indicates that the neural network approach can lead to advances in forecasting snowfall depth.