Comparison of bulk transfer coefficients for heat and
momentum showed that approximations of similarity
functions taken in the WRF might produce erroneous
values of these coefficients. A proposition on iterative
evaluation of the bulk transfer coefficients was suggested.
Behavior of the WRF in the SL when compared
with similarity shows that all the schemes insufficiently
transfer heat from/to the surface, suggesting that diurnal
variation in the WRF SL, and consequently the
whole BL, might be too weak or delayed, especially
for the MYJ scheme. A lack of curvature in the MRF
profiles suggests poor performance of this scheme in
the SL.
Analysis of behavior of the WRF BL schemes above
the SL confirms some of the deficiencies of the schemes
in the SL, primarily insufficient heat transfer for the
MYJ scheme. In the convective conditions with weak
wind, the MRF and YSU develop a BL twice as deep
as the MYJ. In stronger winds (not shown), the YSU
produced a shallower BL, which might signal an improved
performance in light of known overpredictions
of the BL height by the MRF. The MYJ scheme became
numerically unstable for large values of diffusivities
in convective conditions. In neutral conditions
mixing was least pronounced for the YSU scheme, with
the opposite true for the stable BL.
Based on the results of idealized tests, it appears
that the YSU scheme performs better than the MRF
and MYJ under a variety of stability conditions when
compared with theory, LES simulations, and a limited
number of observations. Nevertheless, the conclusions
derived from these tests cannot substitute for extensive
verification of the schemes in daily forecasts evaluated
against a broad range of measurements.