The dependence of QPF on the choice of boundary- and surface-layer parameterization for a lake-effect snowstorm

Six forecasts of a lake-effect snow (LESN) event off Lake Erie were conducted using the Weather Research and Forecasting (WRF) model to determine how the quantitative precipitation forecast (QPF) was affected when the boundary- and surface-layer parameterization schemes were changed. These forecasts showed rather strong variability, with differences in liquid-equivalent precipitation maxima in excess of 20 mm over a six-hour period. The Quasi-Normal Scale Elimination (QNSE) schemes produced the highest accumulations while the Mellor-Yamada-Nakanishi-Niino (MYNN) schemes produced the lowest. Differences in precipitation were primarily due to different sensible heat and moisture fluxes (FH and FQ) off of the lake; with lower FH and FQ in MYNN leading to comparatively weak low-level instability and, consequently, reduced ascent and production of hydrometeors. The different FH and FQ were found to have two causes: In QNSE, the higher FH and FQ were due to the choice to use a Prandtl number (PR) of 0.72 (all other schemes use a PR of 1). In MYNN, the lower FH and FQ were due to the manner in which the similarity stability function for heat (ΨH) is functionally dependent on the temperature gradient between the surface and lowest model layer. It is not known what assumptions are more accurate for environments typical for LESN, but comparisons against available observations and Rapid-Update-Cycle analyses indicated MYNN had the most accurate results.