Ryerson, William, Joshua Hacker, Mary Jordan, and
Kurt Nielsen, Naval Postgraduate School
Skillful fog prediction
relies heavily on skillful NWP predictions of cloud water mixing ratio in the
boundary layer. Motivated by
ceiling and visibility predictions for AFWA, this work evaluates fog-related
predictions from individual members of a 10-member WRF ensemble with 4-km
horizontal resolution. Each member
receives unique initial and lateral boundary conditions and has varied land
surface characteristics, but is also configured with a unique parameterization
suite. The 3-month wintertime
evaluation period encompasses seven California sites representing distinct
coastal, valley, and mountainous mesoclimates. Results from the lowest model layer (roughly 20 m AGL) show
that all physics combinations produce far too many forecasts of zero cloud
water compared to observations, and that this bias is at the expense of cloud
water values corresponding to visibilities of 1-7 mi, which are commonly
observed but rarely predicted by the WRF runs. Lack of cloud water is shown to be mostly due to a warm bias
that is greatest at night and in the coastal region, and varies in magnitude
depending on the WRF member.
Although cloud water predictions are not produced at the 2 m level,
temperature and relative humidity predictions at 2 m have less bias, albeit
with greater inter-member variability.