Transcript MM5I simulations for NARCCAP

MM5I simulations for
NARCCAP
Raymond Arritt
with material borrowed from William Gutowski
Iowa State University
MM5 Heritage
• Started as a 3-layer hurricane model around 1970.
• Evolved into a general mesoscale model with
improved physics, nonhydrostatic dynamical core,
routines for initialization and postprocessing, etc.
– MM4, mid 1980s. First widely used public version.
– MM5 (nonhydrostatic), mid 1990s.
• MM4 / MM5 was probably the most widely used
mesoscale model of its time. Developed an
extensive support structure and user community.
MM5 Heritage
• Descendants:
– RegCM, developed by Giorgi and colleagues starting
in the late 1980s based on the MM4 hydrostatic
dynamical core with different physics (partly adapted
from the NCAR CCM).
– WRF, developed in the late 1990s using many of the
MM4/5 physics options but different dynamical core.
• MM5 is no longer actively supported but
continues to be used. Performs comparably to
other mesoscale weather/climate models.
MM5 features compared to the
NARCCAP suite of RCMs
• Dynamics: Non-hydrostatic (like 2 others)
• Spectral nudging: No (like 3 others)
• Vertical levels: 23 (others: 18 – 35)
• Lateral “sponge zone”: 15(?) grid points (4 – 15)
• Land model: NOAH, 4 soil layers (like 2 others)
• Convection: Kain-Fritsch2 (unique)
• Cloud microphysics: Dudhia simple ice (unique)
• Boundary layer: Hong-Pan, non-local K (like 1
other)
Simulations completed for
NARCCAP
• Current climate driven by NCEP
Reanalysis-2 ("observations")
• Current and future climates driven by
NCAR CCM3
• In progress: current and future climate
driven by HadCM3
– now appears to be running successfully after
a series of false starts
Temperature
Bias [˚C]
DJF
Spatial RMSE = 2.8˚C
Temperature
Bias [˚C]
JJA
Spatial RMSE = 2.3˚C
Precipitation
Bias [%]
DJF
Spatial RMSE = 1.1 mm/d
?
Precipitation
Bias [%]
JJA
Spatial RMSE =
0.6 mm/d
LLJ?
Winter temperature
change for CCSM
driven RCMs
MM5I
CRCM
CCSM
WRFG
Summer temperature
change
CCSM
Summer temperature change likely is
more strongly affected by RCM
physical parameterizations
MM5I
CRCM
WRFG
Winter precipitation
change
CCSM
MM5I
CRCM
WRFG
Summer precipitation
change
CCSM
MM5I
CRCM
WRFG
Ranked Monthly Precipitation for 10
wettest cold season months – Coastal
California
ECPC
wet
HRM3
dry
Final Remarks
• MM5I tends to fall in the middle of the
NARCCAP RCMs, both compared to
observations and in predicted climate changes.
• Note regional differences, especially in summer.
(This does not mean to simply use the “best”
model for a region.)
• See poster: Future Heat Waves in the Greater
St. Louis Area by Claire Steinweg, William J.
Gutowski Jr., Brandon Fisel, Sho Kawazoe