Model Representation of Freezing and Melting Precipitation: Implications for Winter Weather Forecasting

Abstract

During episodes of sustained moderate or heavy precipitation in conjunction with near-freezing temperatures and weak horizontal temperature advection, the latent heat released (absorbed) by the freezing (melting) of falling precipitation may alter thermal profiles sufficiently to affect the type and amount of freezing or frozen precipitation observed at the surface. Representation of these processes by operational numerical weather prediction models is incomplete; forecaster knowledge of these model limitations can therefore be advantageous during winter weather forecasting. The Eta Model employs a sophisticated land surface model (LSM) to represent physical processes at the lower-atmospheric interface. When considering the thermodynamic effect of melting or freezing precipitation at the surface, it is shown that limitations in the current version of the Eta LSM can contribute to biases in lower-tropospheric temperature forecasts. The Eta LSM determines the precipitation type reaching the surface from the air temperature at the lowest model level; subfreezing (above freezing) temperatures are assumed to correspond to snow (rain) reaching the surface. There is currently no requirement for consistency between the LSM and the Eta grid-scale precipitation scheme. In freezing-rain situations, the lowest model air temperature is typically below freezing, and the Eta LSM will therefore determine that snow is falling. As a result, a cold bias develops that is partly caused by the neglected latent heat release accompanying the freezing of raindrops at the surface. In addition, alterations in surface characteristics caused by erroneous snowfall accumulation in the model may also contribute to temperature biases. In an analogous fashion, warm biases can develop in cases with melting snow and above-freezing air temperatures near the surface (the LSM assumes rain). An example case is presented in which model misrepresentation of freezing rain is hypothesized to have contributed to a lower-tropospheric cold bias. A simple temperature correction, based on the first law of thermodynamics, is applied to lower-tropospheric model temperature forecasts; the neglect of latent heat released by freezing rain in the model is shown to contribute substantially to a cold bias in near-surface temperature forecasts. The development of a spurious snow cover likely exacerbated the bias.