Coupled Model Simulations of Boreal Summer Intraseasonal 3050 day Variability: Validation and Cautio

1
Coupled Model Simulations of Boreal Summer
Intraseasonal (30-50 day) Variability Validation
and Caution on Use of MetricsKenneth R.
Sperber1 and H. Annamalai21PCMDI, Lawrence
Livermore National Laboratory, P.O. Box 808,
L-103, Livermore, CA 94550 (sperber1_at_llnl.gov)2In
ternational Pacific Research Center/SOEST,
University of Hawaii, Honolulu, HI 96822
__________________________________________________
__________________________________________________
__________________________________________________
______________
The latitude-time plot of convective anomalies
averaged over India and/or the Bay of Bengal
longitudes has been an accepted diagnostic for
demonstrating the presence of northward
propagation (e.g., Sikka and Gadgil (1980, MWR,
108, 1840-1853). For the observations and a
subset of the models, shown below are
latitude-time plots of the lag regression of the
respective PCs with 20-100 day filtered OLR (W
m-2) that has been averaged between
71.25oE-83.75oE.
Simulation analysis methodology all models are
treated the same, their 20-100 day filtered OLR
is projected onto the Day 0 pattern from
(Annamalai and Sperber 2005, JAS, 62, 2726-2748)
with the resulting PC time series used for lag
regression analysis
78 of northward propagating events occurred in
conjunction with eastward propagation (Lawrence
and Webster 2002, JAS, 59, 1593-1606)
Goals and Background
Day 0
BSISV northward and eastward propagation of AVHRR
OLR (W m-2) using CsEOF analysis (Annamalai and
Sperber 2005, JAS, 62, 2726-2748)
Data
The results above suggest that the models exhibit
northward propagation of convective anomalies
that is essentially consistent with
observations. However, the data above do not
confirm that the anomalies that develops over the
continental latitudes (poleward of 10oN) arises
as an extension of the eastward propagating MJO.
This is investigated below by showing
longitude-time lag plots of convective anomalies
between 5oN-5oS and 15oN-20oN
The ability of the models to simulate the tilted
convection is demonstrated by calculating the Day
10 lag regression of the respective PCs against
20-100 day filtered OLR (W m-2). The pattern
correlation with the Day 10 CsEOF is also given.
The largest pattern correlation for (g)
ECHAM5/MPI-OM incorrectly indicates that this
model gives the best agreement with observations
in simulating the tilted convective anomalies.
The same incorrect result holds true if one
considers the pattern correlation for the
full-space-time BSISV simulated by the models
(not shown). Thus, caution is warranted in using
the pattern correlation as a metric of model
performance without a physical interpretation of
the data
Atmospheric GCMs do not properly represent the
tilted rainband (mm day-1) that is characteristic
of the BSISV (Waliser et al. 2003, Clim. Dynam.,
21, 423-446)
The CMIP3 and CMIP2 (given in red) Models
Caution is warranted in the use of latitude-time
plots for isolating northward propagation. They
must be used in conjunction with longitude-time
plots to determine if the northward propagating
component is associated with the near-equatorial
eastward propagating convective anomalies, as
observed. The ECHAM4/OPYC model has the most
realistic BSISV. Also crucial to this finding was
the examination of animations of the full BSISV
life-cycle of the 20-100 day filtered OLR
anomalies, snapshots of which are shown below
Discovery of Boreal Summer Intraseasonal
Variability
K. R. Sperber was supported under the auspices of
the U.S. Department of Energy Office of Science,
Climate Change Prediction Program by University
of California Lawrence Livermore National
Laboratory under contract No. W-7405-Eng-48.
UCRL-POST-227735