Vertical structure and physical processes of the MaddenâJulian

Transcription

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ???, XXXX, DOI:10.1002/,
1
2
Vertical structure and physical processes of the
Madden–Julian oscillation: Synthesis and summary
1
Nicholas P. Klingaman, Xianan Jiang,
2,3
Duane Waliser,
1
2,3
4
4
Prince K. Xavier, Jon Petch,
1
Steven J. Woolnough
National Centre for Atmospheric Science
and Department of Meteorology, University
of Reading, Reading, UK
2
Joint Institute for Regional Earth
System Science and Engineering, University
of California, Los Angeles, California, USA
3
Jet Propulsion Laboratory and
California Institute of Technology,
Pasadena, California, USA
4
U. K. Met Office, Exeter, UK
D R A F T
April 1, 2015, 9:40am
D R A F T
X -2
3
4
KLINGAMAN ET AL.: MJO PHYSICAL PROCESSES: SYNTHESIS
Abstract.
The “Vertical structure and physical processes of the Madden–Julian os-
5
cillation (MJO)” project comprises three experiments, designed to evaluate
6
comprehensively the heating, moistening and momentum associated with trop-
7
ical convection in general circulation models (GCMs). We consider here only
8
those GCMs that performed all experiments. Some models display relatively
9
higher or lower MJO fidelity in both initialized hindcasts and climate sim-
10
ulations, while others show considerable variations in fidelity between exper-
11
iments. Fidelity in hindcasts and climate simulations are not meaningfully
12
correlated.
13
The analysis of each experiment led to the development of process-oriented
14
diagnostics, some of which distinguished between GCMs with higher or lower
15
fidelity in that experiment. We select the most discriminating diagnostics and
16
apply them to data from all experiments, where possible, to determine if cor-
17
relations with MJO fidelity hold across scales and GCM states. While nor-
18
malized gross moist stability had a small but statistically significant corre-
19
lation with MJO fidelity in climate simulations, we find no link with fidelity
20
in medium-range hindcasts. Similarly, there is no association between timestep-
21
to-timestep rainfall variability, identified from short hindcasts, and fidelity
22
in medium-range hindcasts or climate simulations. Two metrics that relate
23
precipitation to free-tropospheric moisture—the relative humidity for extreme
24
daily precipitation, and variations in the height and amplitude of moisten-
25
ing with rain rate—successfully distinguish between higher- and lower-fidelity
DRAFT
April 1, 2015, 9:40am
DRAFT
26
GCMs in hindcasts and climate simulations. To improve the MJO, develop-
27
ers should focus on relationships between convection and both total mois-
28
ture and its rate of change. We conclude by offering recommendations for
29
further experiments.
D R A F T
April 1, 2015, 9:40am
X-3
D R A F T
X-4
1. Introduction
30
Many contemporary general circulation models (GCMs) used for numerical weather
31
prediction and climate simulations fail to capture the defining characteristics of the
32
Madden–Julian oscillation [MJO; Madden and Julian, 1971, 1972]: its 30–70 day pe-
33
riod, the zonal wavenumber-1 structure of its circulation in the deep tropics and the
34
approximately 5 m s−1 eastward propagation speed of convective anomalies from the In-
35
dian Ocean through the Maritime Continent to the West Pacific [e.g., Lin et al., 2008;
36
Hung et al., 2013]. Zhang [2005] reviews the MJO and its global teleconnections. The
37
“Vertical structure and physical processes of the Madden–Julian oscillation” global-model
38
evaluation project seeks to explore the underlying causes of these deficiencies, which are
39
commonly thought to lie in GCM sub-grid physical parameterizations and the interactions
40
between those parameterizations and the resolved GCM dynamics. The project is orga-
41
nized and supported by the WCRP-WWRP/THORPEX-YOTC MJO Task Force and the
42
GASS panel of GEWEX1 . The overall aim of the project is to characterize, compare and
43
evaluate the heating and moistening processes associated with the the MJO in GCMs,
44
with a particular focus on the vertical structures of those processes [Petch et al., 2011].
45
Such detailed evaluations of simulated processes are possible only because of the substan-
46
tial quantity of satellite-derived observations and high-resolution analyses collected during
47
YOTC [Moncrieff et al., 2012; Waliser et al., 2012].
48
The project comprises three experiments, designed to take advantage of links between
49
GCM biases in initialized hindcasts and those in free-running, multi-decadal climate simu-
50
lations. These links have been demonstrated in previous model inter-comparison projects,
D R A F T
April 1, 2015, 9:40am
D R A F T
X-5
51
such as the Transpose Atmospheric Model Intercomparison Project II [Transpose-AMIP2;
52
Williams et al., 2013] as well as separate efforts by individual groups [e.g., Boyle et al.,
53
2008]. The three experiments are (a) multi-decadal, present-day control climate simula-
54
tions (“20-year climate simulations”); (b) very short initialized hindcasts of two strong,
55
well-observed MJO events in boreal winter 2009–10 (“2-day hindcasts”); and (c) medium-
56
range hindcasts of the same MJO events with a greater range of initial dates (“20-day
57
hindcasts”). Each component collected frequent output, ranging from timestep data for
58
(b) to 6-hr data for (a), of tropospheric vertical profiles of prognostic variables and tenden-
59
cies from individual sub-gridscale paramterizations, to create a rich and comprehensive
60
dataset spanning spatial and temporal scales. We describe and analyze these experiments
61
in separate manuscripts: Jiang et al. [2015], Xavier et al. [2015] and Klingaman et al.
62
[2015] for (a), (b) and (c) above, respectively. Some details on the experiments and their
63
objectives are given in section 2.1, but the reader is encouraged to refer to the above
64
studies.
65
Since the project emphasized links between GCM behavior at short and long temporal
66
scales, some additional analysis is warranted to extend the findings from each experiment.
67
In particular, Jiang et al. [2015] and Klingaman et al. [2015] each identified “process-
68
oriented” diagnostics that the authors found were able to distinguish between GCMs that
69
produced relatively more or less accurate representations of the MJO; hereafter, we refer
70
to this as “MJO fidelity”. GCMs with high MJO fidelity in 20-year climate simulations
71
exhibited strong sensitivity of precipitation to free-tropospheric relative humidity, as well
72
as a tendency for a positive feedback between anomalies in convection and moist static
73
energy via convection-induced vertical circulations. GCMs with high MJO fidelity in
D R A F T
April 1, 2015, 9:40am
D R A F T
X-6
74
the 20-day hindcasts showed a smooth increase in the height of moistening (i.e., positive
75
tendencies of moisture) with precipitation, with low-to-mid-tropospheric moistening at
76
moderate rain rates particularly important. In the 2-day hindcasts, Xavier et al. [2015]
77
demonstrated that GCM biases in temperature and moisture developed quickly, driven by
78
large model-to-model variability in the amount and type of clouds and their interactions
79
with radiation. Several GCMs also showed substantial timestep-to-timestep variability
80
in convection and consequently poor convection–dynamics coupling at the timestep level.
81
From Xavier et al. [2015] alone, however, it was not possible to determine whether these
82
GCM behaviors extended to the other temporal scales captured in the project.
83
Here, we apply the most discriminating and revealing diagnostics developed in each
84
component of the project to the data collected in the other experiments, where possible. Of
85
the 32 GCM contributions received—counting multiple configurations of the same GCM
86
separately—27 GCMs performed climate simulations, 14 performed 20-day hindcasts and
87
12 performed 2-day hindcasts. For consistency, we use only the set of nine GCMs that
88
contributed results to all three experiments (section 2.2). This allows us to achieve our
89
objective of examining GCM behavior across temporal scales, which requires all three
90
experiments, as well as to display and discuss results from all GCMs considered.
91
We have tried to strike a balance between providing the detail needed to understand our
92
diagnostics and repeating information from the other three manuscripts. We give limited
93
information on the data used and calculations performed, such that it should be possible
94
to understand our results using only the information given here, but the reader is strongly
95
encouraged to consider all four manuscripts as a set. We briefly describe the data used in
96
this study (section 2), apply diagnostics developed for one experiment to data from the
D R A F T
April 1, 2015, 9:40am
D R A F T
X-7
97
others (section 3) and discuss (section 4) and summarize (section 5) the results from the
98
project.
2. Models and methods
2.1. Experiments
99
Modeling centers were asked to perform the three experiments with either atmosphere–
100
ocean coupled or atmosphere-only GCMs. The 20-year climate simulations are designed
101
to measure MJO fidelity when the GCM is close to its mean climate; the 2-day hindcasts
102
aim to evaluate parameterization behavior when all GCMs are constrained by a common
103
initial analysis; the 20-day hindcasts bridge the gap between the other experiments by
104
linking hindcast MJO fidelity to biases in physical processes as the GCMs drift towards
105
their preferred states. Hindcasts were initialized daily during two MJO events in YOTC
106
from European Centre for Medium-range Weather Forecasts YOTC (ECMWF-YOTC)
107
00Z analyses: the 20-day (2-day) hindcasts were initialized October 10, 2009–November
108
25, 2009 (October 20, 2009–November 10, 2009) and December 10, 2009–January 25,
109
2010 (December 20, 2009—January 10, 2010). Alongside standard prognostic and surface
110
fields, all experiments obtained tendencies from individual GCM parameterizations and
111
the GCM dynamics for temperature, specific humidity and zonal and meridional winds.
112
The 20-year climate simulations and 20-day hindcasts collected output either globally (for
113
prognostic and surface fields) or across 50◦ S–50◦ N (for tendencies) on a 2.5◦ ×2.5◦ grid
114
and 24 pressure levels; 6-hr (3-hr) data were provided for the 20-year climate simulations
115
(20-day hindcasts). For 2-day hindcasts, centers provided timestep data on the GCM
116
native horizontal and vertical grid over 60◦ –160◦ E, 10◦ S–10◦ N.
D R A F T
April 1, 2015, 9:40am
D R A F T
X-8
2.2. Models
117
We analyse data from the nine GCMs that contributed to all three experiments (Ta-
118
ble 1). In the text, we refer to GCMs by the “Abbreviation” in Table 1; in figures, we
119
label GCMs with the two-letter “Code”. The remainder of this section lists exceptions
120
or caveats to the data and our analysis, to which the reader is encouraged to refer in
121
conjunction with the analysis in section 3.
122
While each center provided a reference for their GCM (Table 1), certain deviations from
123
these or from the experiment design were made that are relevant to our analysis. CanCM4
124
is the only coupled GCM in this set of nine models. To maintain atmosphere–ocean
125
coupled balance, the CanCM4 2-day and 20-day hindcasts were initialized from analyses
126
generated by the operational coupled assimilation system from the the Canadian Centre
127
for Climate Modelling and Analysis, rather than ECMWF-YOTC analyses. Klingaman
128
et al. [2015] and Xavier et al. [2015] discuss the implications for this discrepancy on
129
CanCM4 hindcast fidelity. CAM5 and CAM5-ZM used finite-volume dynamical cores.
130
The GISS-E2’ (“E2 prime”) convection scheme was modified from version E2 to improve
131
tropical intra-seasonal variability [Del Genio et al., 2012; Kim et al., 2012]. For the 2-day
132
and 20-day hindcasts, the choice of SST boundary condition was left to the centers. Among
133
the atmosphere-only models, MRI-AGCM3 and CNRM-AM persisted the initial SST;
134
ECEarth3, MetUM-GA3 and GISS-E2’ persisted the initial SST anomaly with respect to
135
a time-varying climatology; MIROC5, CAM5-ZM and GEOS5 used time-varying observed
136
SSTs. Previous studies have found that using high-frequency, observed SSTs increases
137
MJO prediction skill [e.g., Kim et al., 2008; de Boisséson et al., 2012], but this increase
138
is artifical because it provides the model with information not available at the start of
D R A F T
April 1, 2015, 9:40am
D R A F T
X-9
139
the hindcast. In the set of GCMs used here, Klingaman et al. [2015] found no correlation
140
between the choice of SST boundary condition and hindcast MJO fidelity.
141
For the analysis of moistening tendencies by rain rate (section 3.5), we combine results
142
from two configurations of CAM5: the standard CAM5 and CAM5-ZM, which adds the
143
Song and Zhang [2011] convective microphysics to the standard Morrison and Gettelman
144
[2008] stratiform scheme. We use CAM5-ZM moistening tendencies from the 20-year
145
climate simulations and 20-day hindcasts, as these data were missing or incomplete from
146
CAM5. CAM5-ZM did not perform 2-day hindcasts, so we use CAM5 tendencies instead.
147
Klingaman et al. [2015] and Jiang et al. [2015] demonstrated that CAM5 and CAM5-ZM
148
performed similarly in 20-day hindcasts and 20-year climate simulations, respectively, so
149
this substitution should not affect our results. Further, the data required for the relative
150
humidity difference (section 3.2) and normalized gross moist stability (section 3.3) metrics
151
were not archived from the MetUM-GA3 20-year climate simulation. Although data
152
from the Super-parameterized CAM (SPCAM3) were submitted to all experiments, we
153
exclude SPCAM3 because a different version of the model was used for the 20-year climate
154
simulations than for the 20-day and 2-day hindcasts.
2.3. Data
155
In section 3.5, we compare GCM relationships between moistening tendencies and rain
156
rates to the same relationship in ECMWF-YOTC 24 hour forecasts for the 20-day hind-
157
cast period: October 10, 2009 through February 15, 2010. These short forecasts use the
158
ECMWF Integrated Forecast System (IFS; cycle 35r3) at 16 km horizontal resolution,
159
initialized from the ECMWF-YOTC analyses (also from IFS 35r3) used for the 2-day
160
and 20-day hindcasts. While the IFS parameterizations undoubtedly influence the struc-
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 10
161
ture and amplitude of the moistening tendencies, at these short lead times the model
162
should be reasonably well-constrained at larger scales by the analyses. In the absence of
163
direct observations of moistening, we consider ECMWF-YOTC the closest available ap-
164
proximation to reality. ECMWF-YOTC net moistening was computed by summing the
165
tendencies from the individual IFS physics schemes together with the tendency from the
166
dynamics; this avoids any influence from analysis increments. The 3-hr ECMWF-YOTC
167
data were interpolated to 2.5◦ ×2.5◦ horizontal resolution to agree with data from the
168
20-day hindcasts and 20-year climate simulations.
3. Results
3.1. MJO fidelity in climate and hindcast simulations
169
Jiang et al. [2015] and Klingaman et al. [2015] showed qualitatively that there is lit-
170
tle correspondence between MJO fidelity in the 20-year climate simulations and 20-year
171
hindcasts. Jiang et al. [2015] measured MJO fidelity by pattern correlations of rainfall
172
Hövmoller diagrams between each GCM and observations. The Hövmoller diagrams were
173
constructed from regressions of latitude-averaged (5◦ S–5◦ N), 20–100 day bandpass-filtered
174
precipitation on two base regions: the Indian Ocean (75◦ –85◦ E, 5◦ S-5◦ N) and the West
175
Pacific (130◦ –150◦ E, 5◦ S-5◦ N). The MJO fidelity score is the mean of the two pattern
176
correlation coefficients. In the 20-day hindcasts, Klingaman et al. [2015] assessed MJO
177
fidelity as the first lead time at which the bivariate correlation of the simulated and ob-
178
served Wheeler and Hendon [2004] Real-time Multivariate MJO (RMM) indices was less
179
than a subjectively chosen critical value of 0.7. The RMM skill measure agreed reasonably
180
well, but not perfectly, with pattern correlations of rainfall Hövmoller diagrams from the
181
models, constructed at fixed hindcast lead times from unfiltered daily means, with simi-
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 11
182
larly constructed Hövmollers from TRMM. These pattern correlations approximated the
183
method of Jiang et al. [2015] to the maximum extent possible, given the limited length of
184
the hindcasts.
185
The nine GCMs examined here span the range of fidelity found for all GCMs in the
186
20-day hindcasts and 20-year climate simulations: GISS-E2 and MRI-AGCM3 (CAM5-
187
ZM and GEOS5) were among the highest-fidelity models in the 20-year climate simula-
188
tions (20-day hindcasts), while CanCM4 and MetUM-GA3 (CanCM4 and MIROC5) were
189
among the lowest-fidelity models in the 20-year climate simulations (20-day hindcasts);
190
ECEarth3 and CNRM-AM performed moderately well in both components (Table 2). For
191
these nine GCMs, there is no useful relationship between these fidelity measures (Fig. 1a).
192
Although the two measures are correlated (r=0.55, p∼0.15), if CanCM4 is removed the
193
correlation is weakened substantially (r=0.32, p>0.20) and the regression line becomes
194
nearly vertical. GISS-E2 and MRI-AGCM3 increase in relative performance between the
195
20-day hindcasts and 20-year climate simulations, while CAM5-ZM, GEOS5 and MetUM-
196
GA3 show declines in fidelity; MIROC5, CanCM4, CNRM-AM and ECEarth3 perform
197
similarly in the two experiments, relative to the full set of GCMs in each component. MJO
198
performance in medium-range hindcasts does not necessarily translate to performance in
199
multi-decadal climate simulations, or vice versa, at least not for these GCMs and fidelity
200
measures.
201
In Fig. 1 and throughout, we color models’ codes by their MJO fidelity relative to all
202
GCMs that submitted data for that experiment, not only the nine GCMs shown here.
203
For the 20-year climate simulations, we follow the quartile classifications in Jiang et al.
204
[2015]: the “top 25%” (upper quartile) of models are colored red; the middle 50% are
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 12
205
black; the “bottom 25%” (lowest quartile) of models are blue. For the 20-day hindcasts,
206
we follow the tercile classifications in Klingaman et al. [2015], with the same color scheme
207
as in Jiang et al. [2015]. We note that the classification boundaries do not align between
208
the experiments, but that aligning them would not change our conclusions.
209
Xavier et al. [2015] did not assess MJO fidelity in the 2-day hindcasts, not only because
210
of the limited length and sample size of the hindcasts, but also because the focus of
211
that study was on the short-range, timestep behavior of model physical and dynamical
212
processes, rather than on MJO skill. Here, we attempt to connect MJO fidelity in the 2-day
213
hindcasts to the 20-day hindcasts and 20-year climate simulations by calculating pattern
214
correlations of longitude–time rainfall Hövmoller diagrams between each model at a 2-day
215
lead time (hours 25–48) and TRMM. We construct the Hövmollers as in Klingaman et al.
216
[2015] and compute the pattern correlation over two longitude bands: 60◦ –160◦ E, the full
217
domain of the 2-day hindcasts; and 60◦ –105◦ E, which is approximately the domain of the
218
active MJO phase in the 2-day hindcast cases (see Fig. 1 in Xavier et al. [2015]). We use
219
rainfall data from the 20-day hindcasts for convenience; the 2-day hindcasts are identical
220
to the first two days of the 20-day hindcasts due to the use of the same models and initial
221
conditions. Pattern correlations are computed over all 2-day hindcast start dates for each
222
case, then averaged between the two cases. We also compute the correlations at a 10-day
223
lead time, again using data from the 20-day hindcasts.
224
Table 2 lists the pattern correlations, ranks the models and includes for comparison
225
the classifications of the models in the 20-day hindcasts and 20-year climate simulations.
226
With the exception of CanCM4, all models produce highly similar pattern correlations at
227
2-day lead times for both longitude bands. By this measure of MJO fidelity, there is little
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 13
228
difference among the models at such short lead times. The pattern correlations at 2-day
229
leads show little correspondence to the pattern correlations at 10-day leads: GISS-E2 has
230
a relatively low correlation at 2-day leads, but a relatively higher correlation at 10-day
231
leads; ECEarth3 displays the opposite behavior. Further, the pattern correlations at 2-
232
day leads are poor predictors of MJO fidelity in the 20-day hindcasts and 20-year climate
233
simulations, as shown by the classifications in the right-hand column. Fig. 1b confirms
234
that there is no statistically significant correlation (r=0.27, p>0.20) between the pattern
235
correlations at 2-day and 10-day leads. Similarly, Fig. 1c demonstrates that there is a
236
weak and insignificant correlation (r=0.50, p∼0.15) between the pattern correlations at
237
2-day lead time and MJO fidelity in the 20-year climate simulations. Models that perform
238
similarly well at 2-day lead times, such as CNRM-AM, MIROC5 and CAM5-ZM, show
239
highly disparate fidelity in 20-day hindcasts and 20-year climate simulations. We discuss
240
hypotheses for the disconnects in estimated MJO fidelity among the three experiments in
241
section 4. Because of the similarity in correlation values and the very limited sample of
242
start dates available, we do not separate the 2-day hindcasts based on MJO fidelity; in
243
the figures, all model codes are colored black for relatively “moderate” fidelity.
3.2. Relationship between precipitation and relative humidity
244
Using eight GCMs, Maloney et al. [2014] demonstrated a positive correlation between fi-
245
delity in East Pacific sub-seasonal variability and the difference in average lower-to-middle
246
(850–500 hPa) tropospheric relative humidity (RH) for the heaviest and lightest daily rain
247
rates. This metric measures the sensitivity of simulated precipitation to moisture; various
248
alternate forms have been proposed recently by the MJO community [e.g., Thayer-Calder
249
and Randall , 2009; Xavier , 2012; Kim et al., 2014]. For each of the 27 GCMs for which
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 14
250
20-year climate simulations were submitted, Jiang et al. [2015] computed the difference in
251
mean mass-weighted 850–500 hPa RH for the heaviest 5% and lightest 10% daily rain rates
252
across a Warm Pool domain (60◦ E–180◦ and 15◦ S–15◦ N). The authors found a moderately
253
strong correlation (r=0.45, p∼0.02) between this metric and MJO fidelity.
254
We compute this metric (hereafter “the RH difference metric”) for the 20-day hindcasts
255
from the nine GCMs considered here, combining days 3–20 for all start dates. We exclude
256
the first 48 hours of each hindcast due to large trends in RH as the GCM adjusts its column
257
moisture away from the ECMWF-YOTC analysis. For this reason, we do not compute
258
the RH difference metric for the 2-day hindcasts. We also computed the RH difference
259
metric for days 11–20 only, but found only very small differences (± two percent RH at
260
most) in model scores, suggesting little variation in the metric with lead time after the first
261
two days. There is a moderately strong relationship (r=0.63, p∼0.10) between the RH
262
difference metric and MJO fidelity in the 20-day hindcasts that is statistically significant
263
at the 10% level (Fig. 2a). However, there are five models with similar values of the
264
RH difference metric—GEOS5, MetUM-GA3, GISS-E2, CNRM-AM and MIROC5—but
265
with substantial differences in MJO fidelity in the 20-day hindcasts. While there is a
266
positive overall relationship between the RH difference metric and MJO fidelity in the 20-
267
day hindcasts and 20-year climate simulations, the metric does not discriminate perfectly
268
between higher- and lower-fidelity models.
269
Klingaman et al. [2015] found no statistically significant relationship in the 20-day
270
hindcasts between MJO fidelity and the pattern correlation of specific-humidity anomalies
271
as a function of precipitation rate between each model and ECMWF-YOTC 24-hour
272
forecast data, which is seemingly at odds with our above result for the RH difference
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 15
273
metric. However, Klingaman et al. [2015] considered the full vertical profile of specific-
274
humidity anomalies, as well as all precipitation rates, whereas the RH difference metric
275
focuses on extreme precipitation rates, at both ends of the spectrum, and only 850–
276
500 hPa. We hypothesize that it is the focus on precipitation extremes and the lower
277
troposphere that leads to a stronger connection to MJO fidelity for the RH difference
278
metric than for the specific humidity metric in Klingaman et al. [2015].
279
Next, we compare the RH difference metrics in the 20-day hindcasts and 20-year climate
280
simulations, to assess whether, for a single GCM, variations in the metric can explain
281
variations in MJO fidelity (Fig. 2b). CanCM4 and MIROC5 show relatively low fidelity
282
in both experiments and produce relatively low values of the RH metric. From the 20-day
283
hindcasts to the 20-year climate simulations, CAM5-ZM displays moderate reductions in
284
both MJO fidelity (Fig. 1) and the RH difference metric. Yet GEOS5 also loses fidelity
285
in the 20-year climate simulations relative to the 20-day hindcasts, but has only a slight
286
decrease in the RH difference metric. Of the two models that increase in fidelity between
287
the 20-day hindcasts and 20-year climate simulations, GISS-E2 displays an increase in
288
RH difference while MRI-AGCM3 shows a decline, although MRI-AGCM3 still has one
289
of the highest RH difference metric values among all 20-year climate simulations. The
290
RH difference metric scores in the two experiments are only modestly correlated (r=0.50,
291
p∼0.20) and variations in these scores are only sometimes able to account for variations in
292
MJO fidelity. All GCMs produce RH difference metrics in the 20-year climate simulations
293
that are either less than or similar to their RH difference metrics in the 20-day hindcasts.
294
This suggests that when the GCM mean state is closer to observations, either the GCM
295
precipitation is more sensitive to RH or the GCM dynamic range of RH is greater.
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 16
3.3. Normalized gross moist stability
296
Gross moist stability (GMS) essentially describes how efficiently convection and diver-
297
gent flows remove moisture from an atmospheric column, relative to the import of moisture
298
by advection [Neelin and Held , 1987; Raymond et al., 2009]. Several studies have hypoth-
299
esized that a strong MJO is associated with a negative GMS—a positive feedback in
300
which the active (suppressed) MJO induces a circulation that further moistens (dries) the
301
column—in GCMs and in reality [e.g., Hannah and Maloney, 2011; Benedict et al., 2014;
302
Pritchard , 2014]. Jiang et al. [2015] computed the winter (November–April) normalized
303
GMS (NGMS) over ocean points in an Indo-Pacific domain (60◦ –150◦ E, 15◦ S–15◦ N) fol-
304
lowing the method in Benedict et al. [2014], to which the reader should refer for further
305
details. For all 27 GCMs that performed 20-year climate simulations, there were small
306
but statistically significant negative correlations between MJO fidelity and both vertical
307
NGMS (r=-0.36, p∼0.10) and total NGMS (r=-0.46, p∼0.02).
308
We calculate NGMS from the 20-day hindcasts as in Jiang et al. [2015], using all start
309
dates but only days 3–20, as for the RH difference metric. When we computed NGMS
310
as a function of lead time, by concatenating all hindcast cases at a fixed lead time, most
311
GCMs showed large-amplitude NGMS values—either positive or negative—in the first two
312
days. This suggests strong effects of spin-up on NGMS and prevented us from computing
313
NGMS for the 2-day hindcasts. Since the Benedict et al. [2014] procedure requires a
314
17-day smoothing, we obtain one NGMS value from each 20-day hindcast, using days
315
3–20 and daily means. We then average NGMS across all 94 hindcasts. We stress that
316
this is an extremely small sample of data from which to compute NGMS, which exhibits
317
large variability from one gridpoint and day to the next. NGMS calculations are usually
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 17
318
performed on at least a decade of model data or observations, which span a range of
319
synoptic conditions. We have only 94 days of data, after temporal smoothing, and most
320
of the 20-day hindcast start dates include an active MJO. The results presented below
321
should be taken with caution.
322
In the 20-day hindcasts, we find no useful correlations between MJO fidelity and either
323
vertical (Fig. 3a) or horizontal (Fig. 3c) NGMS, or between fidelity and total NGMS
324
(Fig. 3e). Models with equal fidelity (e.g., MetUM-GA3, GISS-E2, ECEarth3 and MRI-
325
AGCM3) produce NGMS values of opposite signs. The models with the highest (CAM5-
326
ZM) and lowest (CanCM4) MJO fidelity produce similar horizontal and vertical NGMS
327
values. There is no correspondence between vertical NGMS for an individual GCM in the
328
20-day hindcasts and 20-year climate simulations (Fig. 3b), or for total NGMS (Fig. 3f).
329
There is no evidence that, for one GCM, variations in NGMS between the two experiments
330
can account for variations in MJO fidelity. Curiously, there is a negative correlation be-
331
tween horizontal NGMS in the two experiments: GCMs that produce negative horizontal
332
NGMS in 20-day hindcasts (CNRM-AM, CAM5-ZM) show strongly positive horizontal
333
NGMS values in the 20-year climate simulations (Fig. 3d). Further investigation of this
334
behavior is outside the scope of this study, and may be limited to this set of GCMs or
335
caused by the small sample of 20-day hindcast data.
3.4. Timestep precipitation variability
336
In analyzing the 2-day hindcasts, Xavier et al. [2015] discovered substantial timestep-
337
to-timestep intermittency in precipitation in some GCMs, even when gridpoint precipita-
338
tion was averaged across a 5◦ ×5◦ region (75◦ –80◦ E, 0◦ –5◦ N). Timestep intermittency in
339
convection, and hence in heating and moistening increments, could influence the GCM
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 18
340
dynamics and interfere with the propagation of atmospheric waves, including the MJO. In
341
a more detailed analysis of MetUM-GA3, Xavier et al. [2015] showed that the substantial
342
timestep intermittency in convection was not associated with variability in the dynamics,
343
but that this resulted in poor dynamics–convection coupling on the shortest temporal
344
scales. By contrast, MIROC5 had little timestep intermittency in either convection or
345
dynamical fields (e.g., vertical velocity). This analysis did not address whether timestep
346
variability in precipitation affected MJO fidelity.
347
Here, we measure timestep intermittency in precipitation as the lag-1 root-mean-squared
348
difference (RMSD) in timestep precipitation from hours 13–48 of the 2-day hindcasts. Like
349
Xavier et al. [2015], we remove the first 12 hours of each hindcast to limit model spin-up;
350
removing the first 24 hours did not change the conclusions presented below. We compute
351
the lag-1 RMSD from each 2-day hindcast by (a) area-averaging timestep precipitation
352
in 75◦ –80◦ E, 0◦ –5◦ N; (b) creating a second timeseries by lagging the timeseries from (a)
353
by one timestep; (c) computing the RMSD between the timeseries from (a) and (b). We
354
average the RMSD values across all 44 2-day hindcasts.
355
We compare each model’s RMSD values to MJO fidelity in the 20-day hindcasts
356
(Fig. 4a) and 20-year climate simulations (Fig. 4b). While there is a modest nega-
357
tive correlation between the RMSD values and 20-day hindcast fidelity (r=-0.43, p>0.20),
358
this becomes insubstantial if CanCM4 is removed (r=-0.24, p>0.20). Likewise, there is no
359
relationship between the RMSD values and MJO fidelity in the 20-year climate simula-
360
tions (r=0.01, p>0.20). Four models produce low RMSD values—MRI-AGCM3, CAM5,
361
GEOS5 and MIROC5—but vary widely in their fidelity. Likewise, MRI-AGCM3 and
362
GISS-E2 show the highest fidelity, but MRI-AGCM3 generates smooth timestep precip-
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 19
363
itation, while the precipitation in GISS-E2 is highly intermittent. Among these GCMs,
364
there appears to be no relationship between timestep variability in convection and MJO
365
performance. The substantial inter-model variations in timestep precipitation intermit-
366
tency represents an interesting avenue for further research, focusing on the effects on
367
longer temporal scales and interactions with the resolved dynamics.
3.5. Tropospheric moistening
368
Klingaman et al. [2015] found that compositing vertical profiles of net moistening by
369
rain rate produced a diagnostic that was most able to distinguish between higher- and
370
lower-skill GCMs in the 20-day hindcasts, although the distinction was far from absolute.
371
When applied to net moistening and rainfall from ECMWF-YOTC 24-hour forecasts
372
(Fig. 5a), this diagnostic shows that as precipitation increases, the profile of net moist-
373
ening transitions from low-level moistening and upper-level drying at low rain rates (<
374
2 mm day−1 ), through to mid-level moistening at moderate rain rates (2–9 mm day−1 )
375
to upper-level moistening and low-level drying at heavy rain rates (> 9 mm day−1 ), with
376
additional low- and mid-level drying at the strongest rain rates (> 30 mm day−1 ). Pattern
377
correlations of this diagnostic between GCMs and ECMWF-YOTC produced a significant
378
relationship with hindcast skill for all 20-day hindcast GCMs [r=0.82, p∼0.01; Klingaman
379
et al., 2015]. The authors hypothesized that the mid-level moistening at moderate rain
380
rates was critical to a reliable representation of the MJO. In ECMWF-YOTC and the
381
high-skill GCMs, mid-level moistening was produced by a combination of the GCM dy-
382
namics and physics, while in lower-skill GCMs moistening tendencies from the dynamics
383
and physics were nearly always of opposite signs (although not of equal magnitudes).
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 20
384
We compute the composite vertical profiles of net moistening by rain rate (hereafter “the
385
net moistening diagnostic”) using the method in Klingaman et al. [2015]: by composit-
386
ing net moistening (dq/dt), as well as moistening from the GCM dynamics and physics
387
(summing all physics tendencies) within rain-rate ranges over all gridpoints in a Warm
388
Pool domain: 60–160◦ E and 10◦ S–10◦ N, using data from each experiment at the finest
389
temporal and spatial resolutions captured (section 2.1). From the 2-day hindcasts, we
390
use only hours 13–48 of each hindcast and convert the timestep data on the GCM native
391
vertical grid to pressure coordinates using supplied pressure data; we use all days of each
392
20-day hindcast. We aim to understand whether the correlations between MJO fidelity
393
and the net moistening diagnostic hold for the 20-year climate simulations, as well as to
394
assess whether the moistening–rainfall relationships apply to GCM timestep and gridpoint
395
scales.
396
While the net moistening diagnostic produces consistent results in the 20-day hindcasts
397
(middle column of Fig. 5) and 20-year climate simulations (right column) for most GCMs,
398
there are often substantial discrepancies between the 2-day hindcasts (left column) and
399
the other two experiments. For example, in the 2-day hindcasts, CNRM-AM produces a
400
relatively smooth transition between low-level and upper-level moistening (Fig. 5h), but
401
this transition becomes very sharp in the 20-day hindcasts (Fig. 5i) and 20-year climate
402
simulations (Fig. 5j). There is also a marked change in the rain-rate PDF, with many
403
more near-zero values in the timestep PDF from the 2-day hindcasts than in the 3-hr
404
PDF from the 20-day hindcasts. This suggests that the rain-rate PDFs in the 20-day
405
hindcasts and 20-year climate simulations arise from temporal averaging: the PDF peak
406
at 3–5 mm day−1 in the 20-day hindcasts is likely due to averaging several timesteps of
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 21
407
near-zero precipitation together with one timestep of heavy precipitation. As expected,
408
the GCMs that show the largest changes in rain-rate PDFs between the timestep and
409
temporally averaged data—CanCM4, CNRM-AM, GISS-E2 and MetUM—also show the
410
strongest timestep-to-timestep variability in precipitation (Fig. 4). Conversely, GCMs
411
with low timestep variability—CAM5, ECEarth3, MIROC5, MRI-AGCM3 and GEOS5—
412
produce consistent precipitation PDFs at the timestep, 3-hr and 6-hr scales.
413
However, linear combinations of timesteps of high and low rainfall cannot explain many
414
of the variations in the composite net moistening profiles between the 2-day hindcasts and
415
the 20-day hindcasts. Again using CNRM-AM as an example, the strong net drying at
416
750–600 hPa and 2.0–3.0 mm day−1 in the 20-day hindcasts (Fig. 5i) does not appear in
417
the timestep data (Fig. 5h) at that height in any rain-rate band. Similarly, in the 2-day
418
hindcasts CAM5 does not moisten at 700 hPa for any rain rate (Fig. 5b), but CAM5-ZM
419
produces moistening at that level in the 20-day hindcasts (Fig. 5c) and 20-year climate
420
simulations (Fig. 5d). In the 2-day hindcasts, CanCM4 (Fig. 5e), GISS-E2 (Fig. 5q) and
421
MetUM-GA3 (Fig. 5t) display high frequencies of rain rates > 30 mm day−1 associated
422
with very strong column drying, which suggests that the models are still adjusting their
423
moisture fields away from the ECMWF-YOTC analysis. This calls into question the
424
validity of the net moistening diagnostic for the 2-day hindcast data, as well as whether
425
the GCM parameterization behavior seen in the 2-day hindcasts is affected by strong
426
model adjustment away from the ECMWF-YOTC analysis. We obtained similar results
427
for the 2-day hindcasts when we used only hours 25–48. Further, we note that ECEarth3,
428
which is based on the ECMWF IFS, shows little change in either the rain-rate PDF or
429
the net moistening diagnostic between the 2-day hindcasts and the other two experiments
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 22
430
(Fig. 5k–m), supporting the hypothesis that many of the changes in the net moistening
431
diagnostic in other GCMs result from spin-up from a “foreign” analysis.
432
In every GCM, the amplitude of the net moistening diagnostic decreases from the 20-
433
day hindcasts to the 20-year climate simulations. Some of this decrease may be due to
434
temporal averaging, since the 20-day hindcast (20-year climate simulation) diagnostic is
435
computed from 3-hr (6-hr) data, yet averaging the 20-day hindcast data to 6-hr values
436
produced little change in this diagnostic (not shown). The decrease could result from
437
GCM spin-up in the 20-day hindcasts, but Klingaman et al. [2015] noted that there
438
was little variation in this diagnostic with lead time. An alternative hypothesis is that
439
most of the 20-day hindcast start dates include a strong MJO in the initial conditions,
440
while most of these GCMs produce a weaker-than-observed MJO in their 20-year climate
441
simulations. Therefore, the amplitude of the moistening diagnostic may be linked to the
442
amplitude of the MJO, or sub-seasonal tropical convective variability generally, in the
443
simulation. Indeed, CAM5-ZM (Fig. 5c,d) and GEOS5 (Fig. 5o,p) show the largest
444
amplitude reductions and are also the two models that show the largest reductions in
445
MJO fidelity between the 20-day hindcasts and 20-year climate simulations (Fig. 1).
446
As in Klingaman et al. [2015], we compute pattern correlations between the net moist-
447
ening diagnostic for ECMWF-YOTC (Fig. 5a) and each GCM, to investigate the rela-
448
tionship between fidelity in this diagnostic and MJO fidelity. We refer to these pattern
449
correlations as the “net moistening metric”. The 2-day hindcasts show a weak and in-
450
significant correlation (r=0.43, p>0.20) between the net moistening metric and the pattern
451
correlation of the 2-day lead-time rainfall Hövmoller diagram with TRMM (Fig. 6a). This
452
is not surprising, given the inter-model similarity in the rainfall Hövmoller pattern correla-
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 23
453
tions (Table 2) and the inter-model variability in the relationship between net moistening
454
and rainfall (Fig. 5). This result suggests that the net moistening metric is not valid
455
so early in the hindcasts, when the model precipitation and moisture fields may still be
456
adjusting to the analysis. There is a strong relationship (r=0.80, p∼0.01) between the net
457
moistening metric and MJO fidelity in the 20-day hindcasts (Fig. 6b), confirming the re-
458
sults of Klingaman et al. [2015]. We find a slightly weaker, but still significant correlation
459
(r=0.69, p∼0.05) in the 20-year climate simulations (Fig. 6c). Critically, no low-fidelity
460
model scores highly in the net moistening metric for either experiment, nor do any of
461
the higher-fidelity models score poorly, although CNRM-AM performs abnormally poorly
462
in the net moistening metric due to its very sharp transition from drying to moistening
463
throughout the free troposphere around 7 mm day−1 (Fig. 5i,j).
464
Fig. 6d demonstrates that GCMs that increase in relative MJO fidelity (against all
465
models in the experiment, denoted by the colored model codes) from the 20-day hindcasts
466
to the 20-year climate simulations (MRI-AGCM3 and GISS-E2) also show an increase in
467
the net moistening metric, while GCMs that decrease in relative MJO fidelity (GEOS5,
468
CAM5-ZM and MetUM3) display a decrease in the net moistening metric. GCMs in
469
which relative fidelity does not change (CanCM4, CNRM-AM, ECEarth3 and MIROC5)
470
show small changes in the net moistening metric, with the possible exception of CanCM4,
471
which produced a very low value in the 20-day hindcasts. This agrees with our earlier
472
qualitative assessment that the GCMs with the largest amplitude decreases in the net
473
moistening diagnostic between the two experiments also displayed large decreases in MJO
474
fidelity. For this set of GCMs, the net moistening metric distinguishes well between GCMs
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 24
475
with relatively higher and lower MJO fidelity; it also explains variations in MJO fidelity
476
between the two experiments.
4. Discussion
477
The RH difference and net moistening metrics emerge as the measures most able to
478
discern between high- and low-fidelity GCMs across initialized hindcasts and climate
479
simulations, although the relationships with MJO fidelity are far from perfect. These
480
metrics highlight the relationship between precipitation and either total free-tropospheric
481
moisture or its time rate of change. This fits with the recent view that the MJO is a
482
moisture-driven mode of the tropical atmosphere, for which either or both the horizon-
483
tal and vertical advection of moist static energy by the convectively driven circulation is
484
critical to the maintenance and propagation of the mode [e.g., Sobel and Maloney, 2013;
485
Sobel et al., 2014; Hsu et al., 2014]. The RH difference metric emphasizes the sensitivity
486
of simulated rainfall to free-tropospheric moisture, as well as the dynamic range of RH in
487
the GCM. GCMs that produce strong convection and heavy precipitation in atmospheric
488
columns that are far from saturation will score poorly; in these GCMs, the suppressed
489
MJO phase often resembles a weakened active phase.
490
The net moistening metric focuses on the height and sign of moisture tendencies for
491
a given rain rate. GCMs that score well produce free-tropospheric drying from subsi-
492
dence when rain rates are low, low-level and mid-level moistening from advection and
493
convective detrainment as rain rates increase, and upper-level moistening from advection
494
and mid- and low-level drying from precipitation at heavy rain rates. As for the RH
495
difference metric, the net moistening metric rewards GCMs that can build and maintain
496
free-tropospheric moisture anomalies, instead of quickly removing them. Although it has
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 25
497
not been conclusively demonstrated here, it is likely that low-fidelity GCMs remove these
498
moisture anomalies and limit their dynamic range of RH through convection, which in
499
dry columns detrains moisture into the free troposphere and in moist columns removes it
500
by precipitation. This may be inferred from the zero-moistening contours for the GCM
501
dynamics and physics in Fig. 5: in the low- and mid-troposphere, the physics tendency is
502
typically of the opposite sign to the net tendency. The exception is in the mid-troposphere
503
at moderate rain rates, where in the high-fidelity GCMs the physics contributes to the
504
net moistening, rather than working against the dynamics; Klingaman et al. [2015] ar-
505
gued that this was the critical component of the net moistening diagnostic. These results
506
agree with the growing list of studies to show that delaying the response of convection to
507
moisture anomalies improves the MJO in GCMs [e.g., Wang and Schlesinger , 1999; Han-
508
nah and Maloney, 2011; Hirons et al., 2012; Benedict and Maloney, 2013; Klingaman and
509
Woolnough, 2014], as well as with studies that have concluded that the transition from
510
shallow to mid-level convection is critical to the representation of the MJO [e.g., Inness
511
et al., 2001; Benedict and Randall , 2009; Woolnough et al., 2010; Cai et al., 2013]. Fi-
512
nally, we note that the RH difference and net moistening metrics focused, independently,
513
on the low-to-mid troposphere (850–500 hPa) as the key region for obtaining the correct
514
relationship between rainfall and moisture.
515
The 2-day hindcasts were designed to explore GCM parameterization behavior when the
516
models were strongly constrained by a realistic initial analysis. This study has revealed
517
that many diagnostics differ significantly from the 2-day hindcasts to the 20-day hindcasts
518
and 20-year climate simulations. For example, there is little correspondence between GCM
519
skill in precipitation at 2-day and 10-day lead times, as measured by the pattern correlation
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 26
520
of rainfall Hövmoller diagrams with TRMM (Fig. 1). Further, using only the first two
521
days of the 20-day hindcasts produced unrealistically large NGMS values, as well as RH
522
difference metric values that differed substantially from values obtained from days 3–20.
523
The net moistening diagnostic and rain-rate PDFs provided further evidence of different
524
behavior in the 2-day hindcasts, even when only hours 25–48 were examined. While such
525
differences are to be expected, and were the reason for focusing on timestep behavior soon
526
after initialization, it does lead to wonder how these diagnostics evolve as the models drift;
527
it also highlights the challenges in linking the 2-day hindcasts to the 20-day hindcasts and
528
20-year climate simulations. The large differences in some diagnostics may also suggest
529
that we need to investigate further whether, at 48 hours after initialization, the models
530
are still suffering the “shock” of an alien analysis. To help understand the links between
531
the 2-day hindcasts and longer periods, future projects of this nature should also obtain
532
timestep data from a selected period during the medium-range hindcasts (at longer leads,
533
e.g., days 9 and 10) and a short period of the climate simulations. Subsequent projects
534
may also consider performing a parallel set of hindcasts in which models are initialized
535
from their own analysis, although this would restrict participation to modeling centers
536
that have an assimilation system.
537
For many of the nine models considered here, MJO fidelity varies considerably among
538
the 2-day hindcasts, 20-day hindcasts and 20-year climate simulations (Fig. 1). One hy-
539
pothesis for these variations is that there is no single MJO fidelity measure that can be
540
applied to all three experiments, which makes it impossible to cleanly compare fidelity
541
between the temporal scales and simulation types included here. The hindcast simulations
542
are too short to permit the sub-seasonal filtering that is applied so often to isolate the
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 27
543
MJO in fidelity metrics for climate simulations. While the Wheeler and Hendon [2004]
544
RMM indices are often used to evaluate MJO fidelity in initialized hindcasts, there is no
545
accepted RMM-based MJO metric for climate simulations. Many metrics applied to cli-
546
mate simulations identify the MJO with respect to the model’s own mean climate, whereas
547
the metrics applied to hindcasts assess how well the model reproduces the observed MJO.
548
We might have classified the models differently if we had computed MJO fidelity in the
549
20-day hindcasts by projecting each model onto its own leading sub-seasonal empirical
550
orthogonal functions (EOFs), defined from the 20-year climate simulations, rather than
551
the observed EOFs from Wheeler and Hendon [2004]. However, several of the models
552
that performed 20-day hindcasts did not perform 20-year climate simulations. Further,
553
one of the objectives 20-day hindcasts was to identify how well the models predicted the
554
observed MJO, not the MJO from the model’s own mean climate, so that degradations
555
in prediction skill with lead time could be connected to the growth of biases in simulated
556
physical processes.
557
A second hypothesis for the variations in MJO fidelity between the experiments is that
558
hindcasts of two MJO events do not adequately sample model behavior in predicting the
559
MJO, which leads to a biased quantification of prediction skill. There is a strong MJO
560
in all (most) of the initial conditions for the 2-day (20-day) hindcasts. A model that is
561
capable of propagating the initial MJO at roughly the observed phase speed, while limiting
562
drift from the analysis to its own mean climate, will perform well in the 2-day and 20-day
563
hindcasts. Yet to perform well in the 20-year climate simulations, a model must also be
564
able to generate an MJO from quiescent conditions. Assessing skill in MJO genesis for
565
these cases would have required a much broader set of hindcast start dates, which would
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 28
566
have greatly increased the already-large data burden of the 20-day hindcast experiments.
567
We believe a fruitful line of further research exists in the connection between a model’s
568
ability to generate an MJO “from scratch” in initialized hindcasts and the same model’s
569
MJO fidelity in a long climate simulation.
570
We found the 20-day hindcasts were an inconvenient length. They were too short to
571
allow some GCMs to drift fully to their intrinsic climatologies, evidenced by the differences
572
in MJO fidelity between the 20-day hindcasts and 20-year climate simulations (Fig. 1),
573
as well as by comparisons of the climatology of the 20-day hindcasts to the mean seasonal
574
cycle of the 20-year climate simulations at the same time of year (not shown). Yet, the
575
hindcasts were more than long enough to distinguish the higher- and lower-fidelity GCMs,
576
as shown by RMM bi-variate correlations against observations as a function of lead time
577
(see Fig. 2a in Klingaman et al. [2015]). The volume of data requested constrained the
578
sample of MJO events (two) and the number of start dates per case (47). The connection
579
between GCM performance in hindcast and climate simulations would have been improved
580
if we had either (a) more start dates per case, particularly if those start dates did not
581
contain an active MJO, so that we could test the ability of the GCMs to generate an
582
MJO when one was not present in the initial conditions; or (b) hindcasts for more MJO
583
cases, so that we could increase our confidence that GCM performance in the hindcasts
584
was representative of the overall GCM skill. If future projects wish to fully examine
585
model drift, they should use hindcasts of longer than 20 days; 30–35 days would likely be
586
sufficient. If such projects aim to more strongly link GCM fidelity in initialized hindcasts
587
and climate simulations, we suggest that they use shorter hindcasts (e.g., 12 days), but a
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 29
588
wider range of cases and start dates, including many where the initial RMM amplitude is
589
close to zero, as well as multiple ensemble members.
590
Even with the substantial quantity of data collected in YOTC, it is clear that we lack
591
the high-frequency, high-resolution, spatially comprehensive observations necessary to val-
592
idate many of the GCM processes analyzed in this project. This is particularly true at
593
the GCM timestep and gridpoint level, as in the 2-day hindcasts, but also applies to the
594
RH difference and net moistening metrics. While recent campaigns such as the Dynamics
595
of the MJO [DYNAMO; Yoneyama et al., 2013] have collected high-quality observations
596
of diabatic heating and moistening, the short record of these measurements limits their
597
applicability for GCM development, much in the same way as the limited sample of 20-day
598
hindcasts in this project has limited our ability to connect their behavior to the 20-year
599
simulations. It is difficult to interpret process-oriented diagnostics when applied to short
600
datasets, whether those datasets come from models or observations. For the net moist-
601
ening metric, we compared the GCMs to ECMWF-YOTC 24-hour forecasts because of
602
a lack of comprehensive observations. Those moistening tendencies are fundamentally a
603
model-derived product, even though that model is strongly constrained by its own high-
604
quality analysis. We are fortunate that the tendencies come from a model with a realistic
605
MJO, both in this project [Klingaman et al., 2015] and since 2008 generally [e.g., Vitart,
606
2014]. Jiang et al. [2015] compared the RH difference metric in GCMs to TRMM pre-
607
cipitation and ECMWF Interim Reanalysis (ERA-Interim) RH, but TRMM has known
608
issues with detecting light rainfall [e.g., Huffman et al., 2007; Chen et al., 2013] and re-
609
analysis RH is not a substitute for observations. Meaningful progress in improving GCM
610
parameterizations of the diabatic heating, moistening and momentum mixing by tropical
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 30
611
convection requires obtaining high-quality, high-resolution, comprehensive observations
612
of these processes. Specifically, the conclusions of this model-evaluation project advo-
613
cate for high-frequency (i.e., a frequency of an hour or less) observations of precipitation
614
and moistening profiles. These observations must be spatially comprehensive—preferably
615
globally sampled, but at least tropics-wide—and reliable in both clear-sky and cloudy
616
conditions, to adequately capture the transition MJO phase that our results show may be
617
critical for representing the MJO in models.
618
Analysis of observations of diabatic moistening from DYNAMO have shown that there
619
is a strong diurnal cycle in lower-tropospheric moistening during the suppressed phase
620
of the MJO, with a peak moistening in the afternoon following the peak insolation and
621
the diurnal SST maximum [Ruppert and Johnson, 2015]. This moistening is driven by
622
shallow and mid-level convection, but produces little precipitation, presumably because
623
the moisture is detrained quickly into the lower-troposphere. It has been hypothesized
624
that this convection may “recharge” tropospheric moisture, priming the atmosphere for
625
the next MJO active phase. Combined with our results that suggest that capturing the
626
relationship between precipitation and moisture is important for the representation of
627
the MJO in models, these observations provide motivation for a model-evaluation project
628
focused on the MJO events observed during DYNAMO. We recommend an initialized-
629
hindcast experiment for the DYNAMO MJO events, following the suggestions detailed
630
above concerning the length of the hindcasts, the choice of start dates and the collection
631
of timestep data at several points in the hindcasts. The DYNAMO hindcast experiment
632
should focus on the diurnal cycle of convection and on validating models against the
633
wealth of observations of diabatic processes collected in DYNAMO. In addition to a set of
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 31
634
hindcasts initialized from ECMWF analyses, we recommend that modeling centers that
635
have an analysis system be encouraged to perform hindcasts with the model initialized
636
from its own analysis, to quantify the effects of initializing from the “foreign” ECMWF
637
analysis. In our specification for the 20-day and 2-day hindcasts, we overlooked the
638
specification of the SST boundary condition; although we believe that discrepancies in the
639
SST specification did not affect our conclusions, we recommend that any future experiment
640
specify either persisted initial SSTs or persisted initial SST anomalies on a time-varying
641
climatology.
5. Summary and conclusions
642
The “Vertical structure and physical processes of the Madden–Julian oscillation” project
643
has collected and evaluated an extensive dataset of output from 32 GCMs, including tem-
644
perature, moisture and momentum tendencies from individual sub-gridscale parameter-
645
izations. Three experiments were performed that spanned short-range hindcasts, from
646
which timestep output was collected, to long climate simulations with sub-daily data
647
(section 2.1). The results of those experiments are presented in companion manuscripts
648
[Jiang et al., 2015; Xavier et al., 2015; Klingaman et al., 2015], in which diagnostics and
649
metrics are developed that isolate GCM processes (e.g., diabatic heating and moistening
650
associated with tropical convection) to compare GCMs to one another and to observa-
651
tions, where observations exist. In Jiang et al. [2015] and Klingaman et al. [2015], these
652
diagnostics were also correlated against measures of MJO fidelity, but only for the ex-
653
periment considered in each study. Since GCM behavior across temporal scales (i.e.,
654
between timesteps and 6-hr averages) and across model background states (i.e., between
655
initialized hindcasts and free-running climate simulations) is a key focus of the project,
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 32
656
this manuscript has applied the most discerning diagnostics from each component of the
657
project to data from the set of nine GCMs that contributed to all three experiments.
658
We find weak and statistically insignificant relationships between MJO fidelity in the
659
2-day hindcasts, 20-day hindcasts and 20-year climate simulations (Fig. 1). In the 2-
660
day hindcasts, most models show a similar ability to predict the longitude–time pattern
661
of rainfall at a 2-day lead time. The fidelity of these predictions are poorly correlated
662
with fidelity in either the 20-day hindcasts or 20-year climate simulations. Model skill in
663
predicting the Wheeler and Hendon [2004] RMM indices in the 20-day hindcasts is also
664
not significantly correlated with MJO fidelity in the 20-year climate simulations. However,
665
we emphasize that these comparisons of MJO fidelity are far from clean, because we lack
666
a single MJO fidelity metric that can be applied identically in all three experiments.
667
Additionally, fidelity in these initialized hindcasts does not require a model to be able to
668
generate an MJO from quiescent conditions, an ability that is necessary to achieve high
669
fidelity in the climate simulations.
670
In the 20-day hindcasts and 20-year climate simulations, higher-fidelity GCMs tend
671
to score well in the RH difference metric from Jiang et al. [2015] (Fig. 2) and the net
672
moistening metric from Klingaman et al. [2015] (Fig. 6), while low-fidelity GCMs tend
673
to score poorly. However, these relationships are far from perfect, as exemplified by the
674
five GCMs that have similar values of the RH difference metric in the 20-day hindcasts,
675
but which vary considerably in MJO fidelity (Fig. 2a). In most cases, these metrics
676
can account for differences in relative MJO fidelity (i.e., fidelity relative to all GCMs in
677
that experiment, not only the nine considered here) between the experiments: GCMs
678
that increase (decrease) in relative fidelity from one experiment to the other also show an
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 33
679
increase (decrease) in the metric. These results hold more strongly for the net moistening
680
metric, for which correlations with fidelity are higher, but it is difficult to draw conclusions
681
about the relative worth of these two metrics based on only nine GCMs. These results
682
suggest that to improve the MJO, GCM developers should target the relationship between
683
tropical convection and low-to-mid tropospheric moisture, including diabatic moistening
684
by convection, rather than the relationship with diabatic heating, which was not associated
685
with MJO fidelity in the 20-day hindcasts [Klingaman et al., 2015]. To test this hypothesis,
686
we recommend a further initialized-hindcast experiment, focused on the DYNAMO MJO
687
cases, for which detailed observations of diabatic processes are available.
688
Despite a negative correlation between both vertical and total NGMS and MJO fidelity
689
in the 20-year climate simulations, we found no correlations between any component of
690
NGMS and MJO fidelity in the 20-day hindcasts (Fig. 3). Low correspondence between
691
NGMS in the hindcasts and climate simulations may indicate issues with computing
692
NGMS from the limited sample of hindcast data, however. There was also no correlation
693
between MJO fidelity in the 20-day hindcasts and the timestep-to-timestep variability
694
in convection identified by Xavier et al. [2015] in the 2-day hindcasts (Fig. 4). It was
695
often not possible to apply diagnostics developed for the 20-year climate simulations
696
and 20-day hindcasts to the 2-day hindcasts, due to the short record lengths. When
697
diagnostics were applied, such as the net moistening diagnostic, the results proved difficult
698
to interpret in the context of the other two experiments because of strong drift away from
699
the ECMWF-YOTC analyses. The large changes in parameterization behavior across
700
timescales highlights the challenges of linking the behavior of the GCMs when constrained
701
by an analysis to their behavior as they drift towards their intrinsic climatology.
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 34
702
Finally, we note that the complete dataset is available through http://earthsystemcog.org/projects/gass-
703
yotc-mip. While the 2-day hindcast data was archived only over a limited Warm Pool
704
domain, data for the other experiments were collected for at least 50◦ S–50◦ N. We hope
705
that this highly detailed dataset will be useful for a variety of tropical and extra-tropical
706
applications beyond analysis of the MJO.
707
Acknowledgments.
708
All data from all experiments in this project are freely available through
709
http://earthsystemcog.org/projects/gass-yotc-mip. The authors express their gratitude
710
to the nine modeling centers that contributed GCM data to all three experiments in this
711
project, each of which was demanding enough on its own. Comments from Prof. Chi-
712
dong Zhang and an anonymous reviewer helped us to improve the manuscript. Nicholas
713
Klingaman and Steven Woolnough were funded by the National Centre for Atmospheric
714
Science, a collaborative centre of the Natural Environment Research Council, under con-
715
tract R8/H12/83/001. Xianan Jiang acknowledges support by NSF Climate and Large-
716
Scale Dynamics Program under Award AGS-1228302, and NOAA MAPP program under
717
Award NA12OAR4310075. Duane Waliser acknowledges support from the Office of Naval
718
Research under Project ONRBAA12-001, and NSF AGS-1221013, and the Jet Propul-
719
sion Laboratory, California Institute of Technology, under a contract with the National
720
Aeronautics and Space Administration.
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 35
Notes
1. The acronyms refer to the World Climate Research Programme (WCRP), the World Weather Research Programme
(WWRP), The Observing System Research and Predictability Experiment (THORPEX), the Years of Tropical Convection (YOTC), the Global Atmospheric Systems Studies (GASS) panel and the Global Energy and Water Exchanges
Project (GEWEX). At the time of this study, the MJO Task Force was under the collective auspices of WCRP, WWRP,
THORPEX and YOTC. It is currently under the auspices of the Working Group on Numerical Experimentation (WGNE).
721
References
722
723
724
725
Benedict, J. J., and E. D. Maloney (2013), Tropical intraseasonal variability in version 3
of the GFDL atmosphere model, J. Climate, 26, 426–449.
Benedict, J. J., and D. A. Randall (2009), Structure of the Madden–Julian oscillation in
the superparameterized CAM, J. Climate, 66, 3277–3296.
726
Benedict, J. J., E. D. Maloney, A. H. Sobel, and D. M. Frierson (2014), Gross moist
727
stability and MJO simulation in three full-physics GCMs, J. Atmos. Sci., 71, 3327–
728
3349.
729
730
731
732
Boyle, J. S., S. Klein, G. Zhang, S. Xie, and X. Wei (2008), Climate model forecast
experiments for TOGA COARE, Mon. Wea. Rev., 136, 808–832.
Cai, Q., G. J. Zhang, and T. Zhou (2013), Impacts of shallow convection on MJO simulation: a moist static energy and moisture budget analysis, J. Climate, 26, 2417–2431.
733
Chen, Y., E. E. Ebert, K. J. E. Walsh, and N. E. Davidson (2013), Evaluation of TRMM
734
3B42 precipitation estimates of tropical cyclone rainfall using PACRAIN data, J. Geo-
735
phys. Res., 118, 1–13.
736
de Boisséson, E., M. A. Balmaseda, F. Vitart, and K. Mogensen (2012), Impact of the
737
sea surface temperature forcing on hindcasts of Madden–Julian oscillation events using
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 36
738
the ECMWF model, Ocean Sci., 8, 1071–1084.
739
Del Genio, A. D., Y. Chen, Y. Kim, and Y. Mao-Sung (2012), The MJO transition from
740
shallow to deep convection in CloudSat/CALIPSO data and GISS GCM simulations,
741
J. Climate, 25, 3755–3770.
742
743
Hannah, W. M., and E. D. Maloney (2011), The role of moisture–convection feedbacks in
simulating the Madden–Julian oscillation, J. Climate, 24, 2754–2770.
744
Hazeleger, W., X. Wang, C. Severijns, S. Stefanescu, R. Bintanja, A. Sterl, K. Wyser,
745
T. Semmler, S. Yang, B. van den Hurk, B. van Noije, E. van der Linden, and K. van der
746
Wiel (2012), EC-Earth v2.2: description and validation of a new seamless earth system
747
prediction model, Clim. Dynam., 39, 2611–2629.
748
Hirons, L. C., P. Inness, F. Vitart, and P. Bechtold (2012), Understanding advances in the
749
simulation of intraseasonal variability in the ECMWF model. Part II: The application
750
of process-based diagnostics, Q. J. R. Meteorol. Soc., In press, doi:10.1002/qj.2059.
751
752
Hsu, P.-C., T. Li, and H. Murakami (2014), Moisture asymmetry and MJO eastward
propagation in an aquaplanet general circulation model, J. Climate, 27, 8747–8760.
753
Huffman, G. J., R. F. Adler, D. T. Bolvin, G. Gu, E. J. Nelkin, K. P. Bowman, Y. Hong,
754
E. F. Stocker, and D. B. Wolff (2007), The TRMM multi-satellite precipitation analysis:
755
quasi-global, multi-year, combined-sensor precipitation estimates at fine scale, J. Hy-
756
drometeorol., 8, 38–55.
757
Hung, M.-P., J.-L. Lin, W. Wang, D. Kim, T. Shinoda, and S. J. Weaver (2013), MJO and
758
convectively coupled equatorial waves simulated by CMIP5 climate models, J. Climate,
759
26, 6185–6214.
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 37
760
Inness, P. M., J. M. Slingo, E. Guilyardi, and J. Cole (2001), Organization of tropical
761
convection in a GCM with varying vertical resolution: Implications for the simulation
762
of the Madden–Julian Oscillation., Clim. Dynam., 17, 777–793.
763
Jiang, X., D. E. Waliser, P. K. Xavier, J. Petch, N. P. Klingaman, S. J. Woolnough,
764
B. Guan, G. Bellon, T. Crueger, C. DeMott, C. Hannay, H. Lin, W. Hu, D. Kim, C.-L.
765
Lappen, M.-M. Lu, H.-Y. Ma, T. Miyakawa, J. A. Ridout, S. D. Schubert, J. Scinocca,
766
K.-H. Seo, E. Shindo, X. Song, C. Stan, W.-L. Tseng, W. Wang, T. Wu, K. Wyser, G. J.
767
Zhang, and H. Zhu (2015), Vertical structure and physical processes of the Madden–
768
Julian oscillation: Exploring key model physics in climate simulations, J. Geophys. Res.,
769
submitted.
770
Kim, D., A. H. Sobel, A. D. Del Genio, Y. Chen, S. J. Camargo, M.-S. Yao, M. Kelley,
771
and L. Nazarenko (2012), The tropical subseasonal variability simulated in the NASA
772
GISS general circulation model, J. Climate, 25, 4641–4659.
773
774
Kim, D., J.-S. Kug, and A. H. Sobel (2014), Propagating vs. non-propagating Madden–
Julian oscillation events, J. Climate, 27, 111–125.
775
Kim, H.-M., C. D. Hoyos, P. J. Webster, and I.-S. Kang (2008), Sensitivity of MJO
776
simulation and predictability to sea surface temperature variability, J. Climate, 21,
777
5304–5317.
778
Klingaman, N. P., and S. J. Woolnough (2014), Using a case-study approach to improve
779
the Madden–Julian oscillation in the Hadley Centre model, Q. J. R. Meteorol. Soc.,
780
140, 2491–2505.
781
Klingaman, N. P., S. J. Woolnough, X. Jiang, D. Waliser, J. Petch, M. Caian, C. Han-
782
nay, D. Kim, H.-Y. Ma, W. J. Merryfield, T. Miyakawa, M. Pritchard, J. A. Ridout,
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 38
783
R. Roehrig, E. Shindo, F. Vitart, H. Wang, N. R. Cavanaugh, B. E. Mapes, A. Shelly,
784
and G. J. Zhang (2015), Vertical structure and physics processes of the Madden–Julian
785
oscillation: Linking hindcast fidelity to simulated diabatic heating and moistening,
786
J. Geophys. Res., submitted.
787
Lin, J.-L., K. M. Weickman, G. N. Kiladis, B. E. Mapes, S. D. Schubert, M. J. Suarez,
788
J. T. Bacmeister, and M.-I. Lee (2008), Subseasonal variability associated with Asian
789
summer monsoon simulated by 14 IPCC AR4 coupled GCMs, J. Climate, 21, 4542–
790
4566.
791
792
793
794
795
796
Madden, R. A., and P. R. Julian (1971), Detection of a 40–50 day oscillation in the zonal
wind in the tropical Pacific, J. Atmos. Sci., 28, 702–708.
Madden, R. A., and P. R. Julian (1972), Description of global-scale circulation cells in
the tropics with a 40–50 day period, J. Atmos. Sci., 29, 1109–1123.
Maloney, E. D., X. Jiang, S.-P. Xie, and J. Benedict (2014), Process-oriented diagnosis of
East Pacific warm pool intraseasonal variability, J. Climate, 27, 6305–6324.
797
Merryfield, W. J., W.-S. Lee, G. J. Boer, V. V. Kharin, J. F. Scinocca, G. M. Flato, R. S.
798
Ajayamohan, J. C. Fyfe, Y. Tang, and S. Polavarapu (2013), The Canadian seasonal to
799
interannual prediction system. Part I: Models and initialization, Mon. Wea. Rev., 141,
800
2910–2945.
801
Moncrieff, M. W., D. E. Waliser, M. J. Miller, M. A. Shapiro, G. R. Asrar, and J. Caughey
802
(2012), Multiscale convective organisation and the YOTC virtual global field campaign,
803
Bull. Amer. Meteor. Soc., 93, 1171–1187.
804
Morrison, H., and A. Gettelman (2008), A new two-moment bulk stratiform cloud micro-
805
physics scheme in the community atmosphere model, version 3 (cam3), J. Climate, 21,
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 39
806
3642–3659.
807
Neale, R. B., et al. (2012), Description of the NCAR Atmospheric Model: CAM5.0,
808
Tech. Rep. NCAR/TN-486+STR, National Center for Atmospheric Research, Boulder,
809
Colorado, USA.
810
811
Neelin, J. D., and I. M. Held (1987), Modeling tropical convergence based on the moist
static energy budget, Mon. Wea. Rev., 115, 3–12.
812
Petch, J., D. Waliser, X. Jiang, P. K. Xavier, and S. Woolnough (2011), A global model
813
intercomparison of the physical processes associated with the Madden–Julian oscillation,
814
GEWEX News, August, 5.
815
816
817
818
Pritchard, M. S. (2014), Causal evidence that rotational moisture advection is critical to
the superparameterized Madden–Julian oscillation, J. Atmos. Sci., 71, 800–815.
Raymond, D. J., S. S. Sessions, A. H. Sobel, and Z. Fuchs (2009), The mechanics of gross
moist stability, J. Adv. Model. Earth Sys., 1, 1–19.
819
Rienecker, M. M., M. J. Suarez, R. Todling, J. Bacmeister, L. Takacs, H.-C. Liu, W. Gu,
820
M. Sienkiewicz, R. D. Koster, R. Gelaro, I. Stajner, and J. E. Nielsen (2008), The
821
GEOS-5 data assimilation system: Documentation of version 5.0.1, 5.1.0 and 5.2.0,
822
Tech. Rep. series on Global Modeling and Data Assimilation, National Aeronautics and
823
Space Administration.
824
Ruppert, J. H., and R. H. Johnson (2015), Diurnally modulated cumulus moistening in
825
the pre-onset stage of the Madden–Julian oscillation during DYNAMO, J. Atmos. Sci.,
826
accepted, doi:10.1175/JAS–D–14–0218.1.
827
Schmidt, G., M. Kelley, L. Nazarenko, R. Ruedy, G. L. Russell, I. Aleinov, M. Bauer, S. E.
828
Bauer, M. K. Bhat, R. Bleck, V. Canuto, Y.-H. Chen, Y. Cheng, T. L. Clune, A. Del Ge-
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 40
829
nio, R. de Fainchtein, G. Faluvegi, J. E. Hansen, R. J. Healy, N. Kiang, D. Koch, A. A.
830
Lacis, A. N. LeGrande, J. Lerner, K. K. Lo, E. E. Matthews, S. Menon, R. L. Miller,
831
V. Oinas, A. O. Oloso, J. P. Perlwitz, M. J. Puma, W. M. Putman, D. Rind, A. Ro-
832
manou, M. Sato, D. T. Shindell, S. Sun, R. A. Syed, N. Tausnev, K. Tsigaridis, N. Unger,
833
A. Voulgarakis, M.-S. Yao, and J. Zhang (2014), Configuration and assessment of the
834
GISS ModelE2 contributions to the CMIP5 archive, J. Adv. Model. Earth Syst., 6,
835
141–184.
836
837
838
839
Sobel, A. H., and E. M. Maloney (2013), Moisture modes and the eastward propagation
of the MJO, J. Atmos. Sci., 70, 187–192.
Sobel, A. H., S. Wang, and D. Kim (2014), Moist static energy budget of the MJO during
DYNAMO, J. Atmos. Sci., 71, 4276–4291.
840
Song, X., and G. J. Zhang (2011), Microphysics parameterization for convective clouds in
841
a global climate model: Description and single-column model tests, J. Geophys. Res.,
842
116, D02,201.
843
Song, X., G. J. Zhang, and J.-L. F. Li (2012), Evaluation of microphysics parameterization
844
for convective clouds in the NCAR Community Atmosphere Model CAM5, J. Climate,
845
25, 8568–8590.
846
847
848
849
Thayer-Calder, K., and D. A. Randall (2009), The role of convective moistening in the
Madden–Julian oscillation, J. Atmos. Sci., 66, 3297–3312.
Vitart, F. (2014), Evolution of ECMWF sub-seasonal forecast skill scores, Q. J. R. Meteorol. Soc., 683, 1889–1899.
850
Voldoire, A., E. Sanchez-Gomez, D. Salas y Mlia, B. Decharme, C. Cassou, S. Sénési,
851
S. Valcke, I. Beau, A. Alias, M. Chevallier, M. D´
qué, J. Deshayes, H. Douville, E. Fer-
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 41
852
nandez, G. Madec, E. Maisonnave, M.-P. Moine, S. Planton, D. Saint-Martin, S. Szopa,
853
S. Tyteca, R. Alkama, S. Belamari, A. Braun, L. Coquart, and F. Chauvin (2013), The
854
CNRM-CM5.1 global climate model: description and basic evaluation, Clim. Dynam.,
855
40, 2091–2121.
856
Waliser, D. E., M. W. Moncrieff, D. Burridge, A. H. Fink, D. Gochis, B. N. Goswami,
857
B. Guan, P. Harr, J. Heming, H.-H. Hsu, C. Jakob, M. Janiga, R. Johnson, S. Jones,
858
P. Knippertz, J. Marengo, H. Nguyen, M. Pope, Y. Serra, C. Thorncroft, M. Wheeler,
859
R. Wood, and S. Yuter (2012), The “Year” of Tropical Convection (May 2008–April
860
2010), Bull. Amer. Meteorol. Soc., 93, 1189–1218.
861
Walters, D. N., M. J. Best, A. C. Bushell, D. Copsey, J. M. Edwards, P. D. Falloon, C. M.
862
Harris, A. P. Lock, J. C. Manners, C. J. Morcrette, M. J. Roberts, R. A. Stratton,
863
S. Webster, J. M. Wilkinson, M. R. Willett, I. A. Boutle, P. D. Earnshaw, P. G. Hill,
864
C. MacLachlan, G. M. Martin, W. Moufouma-Okia, M. D. Palmer, J. C. Petch, G. G.
865
Rooney, A. A. Scaife, and K. D. Williams (2011), The Met Office Unified Model Global
866
Atmosphere 3.0/3.1 and JULES Global Land 3.0/3.1 configurations, Geosci. Model
867
Dev., 4, 919–941.
868
Wang, W., and M. E. Schlesinger (1999), The dependence on convective parameterization
869
of the tropical intraseasonal oscillation simulated by the UIUC 11-layer atmospheric
870
GCM, J. Climate, 12, 1423–1457.
871
Watanabe, M., T. Suzuki, R. O´ıshi, Y. Komuro, S. Watanabe, S. Emori, T. Take-
872
mura, M. Chikira, T. Ogura, M. Sekiguchi, K. Takata, D. Yamazaki, T. Yokohata,
873
T. Nozawa, H. Hasumi, H. Tatebe, and M. Kimoto (2010), Improved climate simu-
874
lation by MIROC5: Mean states, variability and climate sensitivity, J. Climate, 23,
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 42
875
6312–6335.
876
Wheeler, M. C., and H. H. Hendon (2004), An all-season real-time multivariate MJO
877
index: Development of an index for monitoring and prediction, Mon. Wea. Rev., 132,
878
1917–1932.
879
Williams, K. D., A. Bodas-Salcedo, M. Déqué, S. Fermepin, B. Medeiros, M. Watanabe,
880
C. Jakob, S. A. Klein, C. A. Senior, and D. L. Williamson (2013), The Transpose-AMIP
881
II experiment and its application to the understanding of Southern Ocean cloud biases
882
in climate mdoels, J. Climate, 26, 3258–3274.
883
Woolnough, S. J., P. N. Blossey, K.-M. Xu, P. Bechtold, J.-P. Chaboureau, T. Hosomi,
884
S. F. Iacobellis, Y. Luo, J. C. Petch, R. Y. Wong, and S. Xie (2010), Modelling convec-
885
tive processes during the suppressed phase of a Madden–Julian oscillation: Comparing
886
single-column models with cloud-resolving models, Q. J. R. Meteorol. Soc., 136, 333–
887
353.
888
889
Xavier, P. K. (2012), Intraseasonal convective moistening in CMIP3 models, J. Climate,
25, 2569–2577.
890
Xavier, P. K., J. C. Petch, N. P. Klingaman, S. J. Woolnough, X. Jiang, D. E. Waliser,
891
M. Caian, S. M. Hagos, C. Hannay, D. Kim, J. Cole, T. Miyakawa, M. Prithard,
892
R. Roehrig, E. Shindo, F. Vitart, and H. Wang (2015), Vertical structure and physical
893
processes of the Madden–Julian oscillation: Biases and uncertainties at short range,
894
J. Geophys. Res., submitted.
895
896
Yoneyama, K., C. Zhang, and C. N. Long (2013), Tracking pulses of the Madden–Julian
oscillation, Bull. Amer. Meteorol. Soc., 94, 1871–1891.
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 43
897
Yukimoto, S., Y. Adachi, M. Hosaka, T. Sakami, H. Yoshimura, M. Hirabara, T. Y.
898
Tanaka, E. Shindo, H. Tsujino, M. Deushi, R. Mizuta, S. Yabu, A. Obata, H. Nakano,
899
T. Ose, and A. Kitoh (2012), A new global model model of meteorological research insti-
900
tute: MRI-CGCM3–model description and basic performance, J. Meteorol. Soc. Japan,
901
90A, 23–64.
902
903
Zhang,
C.
(2005),
Madden–Julian
oscillation,
Rev. Geophys.,
43,
RG2003,
doi:10.1029/2004RG000,158.
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 44
b. Rainfall Hövmoller: 10-day lead vs. 2-day lead
0.70
0.95
MR
GI
Correlation of rainfall Hovmoller at 2-day lead
Fidelity in 20-year simulations (correlation of rainfall Hovmoller)
a. Fidelity: 20-day vs. 20-year
MR
0.85
E3
CN
CZ
0.75
NA
r = 0.55
0.65
MI
0.55
MO
E3
0.65
CN
CZ
NA
MI
MO
GI
0.60
0.55
r = 0.27
0.50
CC
CC
0.45
-0.05
0.45
5
8
11
14
17
20
Fidelity in 20-day hindcasts (lead of RMM bivar corr < 0.7; days)
0.05
0.15
0.25
0.35
0.45
c. 20-year fidelity vs. 2-day lead Hövmoller
0.70
MR
NA
0.65
CN
MI
MO
GI
0.60
r = 0.50
0.55
0.50
CC
0.45
0.45
Figure 1.
CZ E3
0.55
0.65
0.75
0.85
Fidelity in 20-year simulations
0.95
For each model, (a) MJO fidelity in 20-day hindcasts against fidelity in 20-year
climate simulations; (b) pattern correlations of Hövmoller diagarms of daily-mean rainfall (averaged 10◦ S–10◦ N and computed over 60◦ –160◦ E) between TRMM and the model, at 2-day and
10-day lead times; (c) fidelity in 20-year climate simulations against the 2-day pattern correlations from (b). In (a), the color of the first letter of each code gives relative fidelity among all 13
20-day hindcast models (blue: lower tercile; black: middle tercile; red: upper tercile); the color
of the second letter gives relative fidelity among all 27 20-year climate simulation models (blue:
lower 25%; black: middle 50%; red: upper 25%). In (b), the codes show 20-day hindcast fidelity;
in (c), they show fidelity in 20-year climate simulations. The least-squares regression lines and
correlation coefficients are also shown. Without CanCM4 (CC) the correlation in (a) is 0.32.
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 45
a. RH difference: 20-day hindcasts
b. RH difference: 20-day vs. 20-year
RH difference in 20-year climate simulations (%)
CZ
Fidelity in 20-day hindcasts (days)
20
NA
17
MO
GI
E3 MR
MI CN
14
r = 0.63
11
8
CC
5
40
45
50
55
RH difference in 20-day hindcasts (%)
Figure 2.
60
60
55
GI
CN
50
MR
E3
CZ
45
CC
NA
r = 0.50
40
MI
35
35
40
45
50
55
RH difference in 20-day hindcasts (%)
60
For each model, the difference in mass-weighted 850–500 hPa averaged relative
humidity between the heaviest 5% and lightest 10% of daily rainfall events in the 20-day hindcasts
(using days 3–20 only), against (a) MJO fidelity in the 20-day hindcasts and (b) the same
difference in relative humidity from the corresponding 20-year climate simulations. The leastsquares regression lines and correlation coefficients are also shown; (b) also includes a one-to-one
line. In (a), model codes are colored by fidelity in 20-day hindcasts only, relative to all 13 20-day
hindcast models. In (b), model codes are colored as in Fig. 1. Required data for this diagnostic
were not archived from the MetUM-GA3 20-year climate simulation, so that model is excluded
from (b).
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 46
a. Vertical NGMS: 20-day hindcasts
b. Vertical NGMS: 20-day vs. 20-year
0.40
CZ
Vertical NGMS in 20-year climate simulations
20
NA
17
GI
MO
E3
MR
CN
MI
r = -0.08
14
11
8
CC
5
-0.30
-0.20
-0.10
0.00
0.10
0.20
Vertical NGMS in 20-day hindcasts
Horizontal NGMS in 20-year climate simulations
20
NA
E3
MR GI
MO
CN
MI
14
r = 0.28
11
8
CC
5
-0.80
-0.60
-0.40
-0.20
0.00
0.20
Horizontal NGMS in 20-day hindcasts
Total NGMS in 20-year climate simulations
NA
MR
GI
MO
MI
14
r = 0.26
11
8
CC
5
-0.80
MR
r = 0.28
-0.10
CN
-0.20
-0.30
0.40
0.80
0.60
0.40
0.20
CN
CZ
E3
NA
MIMR
GI
CC
r = -0.35
0.00
-0.20
-0.60
-0.40
-0.20
0.00
0.20
Horizontal NGMS in 20-day hindcasts
0.40
0.80
20
E3
CZ
GI
0.00
f. Total NGMS: 20-day vs. 20-year
CZ
CN
0.10
-0.40
-0.80
0.40
e. Total NGMS: 20-day hindcasts
17
NA E3
d. Horizontal NGMS: 20-day vs. 20-year
CZ
17
MI
0.20
-0.40
-0.40 -0.30 -0.20 -0.10 0.00 0.10 0.20 0.30
Vertical NGMS in 20-day hindcasts
0.30
c. Horizontal NGMS: 20-day hindcasts
CC
0.30
-0.60 -0.40 -0.20 0.00 0.20 0.40
Total NGMS in 20-day hindcasts
0.60
0.60
CC
0.40
CZ
MI
NA
E3
0.20
CN
MR
GI
r = -0.06
0.00
-0.20
-0.40
-0.80
-0.60
-0.40
-0.20
0.00
0.20
Total NGMS in 20-day hindcasts
0.40
Figure 3. For (a,b) vertical normalized gross moist stability (NGMS), (c,d) horizontal NGMS
and (e,f) total NGMS: (a,c,e) MJO fidelity in 20-day hindcasts against NGMS; (b,d,f) NGMS
in 20-day hindcasts against NGMS in 20-year climate simulations. The least-squares regression
lines and correlation coefficients are also shown; (b,d,f) also include a one-to-one line. In (a,c,e),
model codes are colored by fidelity in 20-day hindcasts, as in Fig. 2a; in (b,d,f), model codes are
colored as in Fig. 1a. For the 20-day hindcasts, NGMS is computed from only days 3–20.
D R A F T
April 1, 2015, 9:40am
D R A F T
a. Against fidelity in 20-day hindcasts
b. Against fidelity in 20-year climate simulations
1.00
C5
Fidelity in 20-year climate simulations (days)
20
NA
17
MR
E3
MO
GI
CN
14
X - 47
MI
r = -0.43
11
8
CC
GI
0.90
MR
E3
0.80
0.70
CN
C5
r = 0.014
NA
0.60
MI
MO
CC
0.50
0.40
5
0
1
2
3
4
5
6
7
8
9
RMSD in timestep rainfall in 2-day hindcasts (mm day-1)
Figure 4.
0
1
2
3
4
5
6
7
8
9
RMSD in timestep rainfall in 2-day hindcasts (mm day-1)
For each model, the lag-1 root mean squared difference (RMSD) in area-averaged
(75◦ –80◦ E, 0◦ –5◦ N) timestep precipitation in the 2-day hindcasts (mm day−1 ) against MJO fidelity in the (a) 20-day hindcasts and (b) 20-year climate simulations. The least-squares regression lines and correlation coefficients are also shown. Without CanCM4 (CC) the correlation in
(a) -0.24. Model codes are colored by fidelity in (a) 20-day hindcasts and (b) 20-year climate
simulations, relative to all GCMs for which 20-day hindcasts and 20-year climate simulations,
respectively, were submitted.
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 48
0
0
0
1.00
0
0.60
0.40
0.30
0.20
0.15
0.10
0
400
500
600
0.06
0
0.04
0.03
800
850
0.02
900
0.015
950
0
0
0
1000
0.01
0 0 0 0 0 0 0 0 0 0 0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12 16 2-10 25 30 > 30
Daily-mean precipitation rate (mm day )
700
Fraction of points
50
100
150
200
250
300
0
Pressure (hPa)
a. ECMWF-YoTC and TRMM
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 -10.3 0.5
0.7 0.9 1.1 1.3 1.5
Total dq/dt (g kg day-1)
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
800
900
0
0
0
0
800
900
0
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
Precipitation rate (mm day )
0.06
0.04
0.03
0.02
0.015
0.01
1.00
0
300
0.60
0.40
0.30
0.20
0.15
0.10
0
400
0
500
600
700
800
900
0
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
Total dq/dt (g kg-1 day-1)
0.06
0.04
0.03
0.02
0.015
0.01
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
h. CNRM-AM 2-day
400
500
0
600
700
0
800
900
0
0
1000
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
0
0
300
0
400
500
600
700
800
900
0
0
1000
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
0.06
0.04
0.03
0.02
0.015
0.01
k. ECEarth3 2-day
400
0
500
600
0
0
700
800
900
0
0
0 0
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
Figure 5.
0
0
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
0
0
1.00
0
100
0.60
0.40
0.30
0.20
0.15
0.10
200
300
0
400
500
600
700
800
900
0
1000
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
m. ECEarth3 20-year
0
0
1.00
0
0.60
0.40
0.30
0.20
0.15
0.10
200
Pressure (hPa)
0.60
0.40
0.30
0.20
0.15
0.10
0
0
900
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
100
Fraction of points
300
700
800
l. ECEarth3 20-day
1.00
200
600
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
100
500
1.00
0.60
0.40
0.30
0.20
0.15
0.10
200
0.10
0.06
0.04
0.03
0.02
0.015
0.01
1.00
0.60
0.40
0.30
0.20
0.15
0
j. CNRM-AM 20-year
0
0
0
0
300
400
1.00
0.60
0.40
0.30
0.20
0.15
0.10
300
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
100
Pressure (hPa)
0.60
0.40
0.30
0.20
0.15
0.10
Fraction of points
1.00
200
0
200
i. CNRM-AM 20-day
100
0
00
0
0
100
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
0.10
0.06
0.04
0.03
0.02
0.015
0.01
1.00
0.60
0.40
0.30
0.20
0.15
Pressure (hPa)
g. CanCM4 20-year
0
Fraction of points
600
700
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
Fraction of points
500
0
200
Pressure (hPa)
400
0
100
Fraction of points
Pressure (hPa)
0
0
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
0
0.60
0.40
0.30
0.20
0.15
0.10
200
300
900
f. CanCM4 20-day
1.00
0
0
700
800
Fraction of points
e. CanCM4 2-day
0
600
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
100
500
300
0
0
400
500
0
600
700
800
900
0
0
0
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
0
0
0
100
300
0
0
400
500
600
0
700
800
900
1.00
0.60
0.40
0.30
0.20
0.15
0.10
200
0
0
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
Fraction of points
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
0.10
0.06
0.04
0.03
0.02
0.015
0.01
1.00
0.60
0.40
0.30
0.20
0.15
Pressure (hPa)
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
Pressure (hPa)
700
0
0
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
600
0
900
0
0
500
0.60
0.40
0.30
0.20
0.15
0.10
0
0
0
800
0.06
0.04
0.03
0.02
0.015
0.01
400
Pressure (hPa)
0
700
Fraction of points
600
300
Pressure (hPa)
500
400
1.00
00
0
200
Pressure (hPa)
0
0
0
100
Fraction of points
400
300
0
Pressure (hPa)
0
0.60
0.40
0.30
0.20
0.15
0.10
200
0
0.60
0.40
0.30
0.20
0.15
0.10
300
Pressure (hPa)
200
d. CAM5-ZM 20-year
1.00
0
0
100
Fraction of points
1.00
Fraction of points
c. CAM5-ZM 20-day
100
Fraction of points
b. CAM5 2-day
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
Shading shows the mean vertical profiles of the rate of change of specific humidity
(dq/dt; g kg day−1 ) for each range of rain rates on the horizontal axis. Solid (dotted) lines are
zero contours of the dq/dt from GCM dynamics (physics). Dynamics tendencies are positive
(moistening) above and to the right of the solid line; physics tendencies are positive below and
to the left of the dotted line. Dashed lines show PDFs of rain rates, using the right-hand
vertical axis. Composites are computed from timestep data for 2-day hindcasts, 3-hr data for
20-day hindcasts and 6-hr data for 20-year climate simulations. For 20-day hindcasts and 20-year
climate simulations, colored panel labels indicate MJO fidelity.
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 49
0
0
0
0
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0
400
500
600
0.06
0
0.04
0.03
800
850
0.02
900
0.015
950
0
0
0
1000
0.01
0 0 0 0 0 0 0 0 0 0 0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12 16 2-10 25 30 > 30
Daily-mean precipitation rate (mm day )
700
Fraction of points
50
100
150
200
250
300
0
Pressure (hPa)
a. ECMWF-YoTC and TRMM
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 -10.3 0.5
0.7 0.9 1.1 1.3 1.5
Total dq/dt (g kg day-1)
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
700
800
900
700
800
0
900
0
0
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
300
0
400
500
600
700
800
900
0
0
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
0.06
0.04
0.03
0.02
0.015
0.01
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
1.00
0.60
0.40
0.30
0.20
0.15
0.10
200
0
400
0
0
0
600
700
800
0
900
0
0
1000
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
0
0
300
0
0
400
0
500
600
700
800
0
900
0
0
1000
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
0
400
0
500
0
600
700
800
0
0
900
0
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
0
0
0
0
300
400
0
500
600
700
0
800
900
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
0.06
0.04
0.03
0.02
0.015
0.01
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
z. MRI-AGCM3 2-day
0
0
400
500
600
0
0
700
800
0
900
0
0
0
1000
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
0
0
700
900
0
0
1000
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
0
0
100
300
0
0
400
500
0
600
700
0
800
1.00
0.60
0.40
0.30
0.20
0.15
0.10
200
900
0
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
ab. MRI-AGCM3 20-year
0
0
0
0
300
0
0
200
Pressure (hPa)
0.60
0.40
0.30
0.20
0.15
0.10
300
600
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
100
Fraction of points
200
500
800
aa. MRI-AGCM3 20-day
1.00
0
0
400
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
0.01
0.02
0.015
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
100
300
1.00
0.60
0.40
0.30
0.20
0.15
0.10
200
1.00
0.60
0.40
0.30
0.20
0.15
0.10
200
0
300
0
0
100
y. MIROC5 20-year
0
Pressure (hPa)
0.60
0.40
0.30
0.20
0.15
0.10
0
0
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
100
Fraction of points
200
0
0.06
0.04
0.03
0.02
0.015
0.01
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
x. MIROC5 20-day
1.00
0
900
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
w. MIROC5 2-day
0
0
700
1.00
0.60
0.40
0.30
0.20
0.15
0.10
200
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
0
600
v. MetUM-GA3 20-year
0
0.01
0.02
0.015
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
100
500
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
Fraction of points
500
400
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
100
Pressure (hPa)
0
300
800
u. MetUM-GA3 20-day
Fraction of points
100
0.60
0.40
0.30
0.20
0.15
0.10
200
Fraction of points
t. MetUM-GA3 2-day
1.00
0
0
0
100
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
0.01
0.02
0.015
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
Pressure (hPa)
0.60
0.40
0.30
0.20
0.15
0.10
200
Pressure (hPa)
600
s. GISS-E2 20-year
0
0
500
0
0
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
1.00
0
0
100
Fraction of points
400
300
900
0.06
0.04
0.03
0.02
0.015
0.01
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
0
Pressure (hPa)
300
0.60
0.40
0.30
0.20
0.15
0.10
0
0
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
Fraction of points
1.00
200
700
800
r. GISS-E2 20-day
100
600
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
q. GISS-E2 2-day
Pressure (hPa)
0
500
Fraction of points
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
0.01
0.02
0.015
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
Pressure (hPa)
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
Fraction of points
0
0
600
Fraction of points
900
0
1000
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
500
0
0.60
0.40
0.30
0.20
0.15
0.10
0
0
400
0
0
500
600
700
800
900
0
0
0
0
0
0
1.00
0
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
0.06
0.04
0.03
0.02
0.015
0.01
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
0
0
100
0
0
0
0
0
200
300
0
0
400
0
500
0
600
700
800
0
900
0
0
0
0
0
0
1000
0
0
0
0
0
0
0
0
0
0 .0 .0 .0 .0 .0 .0
0
< 0.1 0.1 0.3 0.6 1.0 1.5 2.0 3.0 5.0 7.0 9.0 12-1 16 20 25 30 > 30
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0.06
0.04
0.03
0.02
0.015
0.01
Fraction of points
800
0.06
0.04
0.03
0.02
0.015
0.01
0
0
0
0.60
0.40
0.30
0.20
0.15
0.10
0
400
Pressure (hPa)
0
700
300
Pressure (hPa)
0
600
0
0
400
Pressure (hPa)
500
300
200
Pressure (hPa)
0
1.00
100
Fraction of points
0
1.00
0.60
0.40
0.30
0.20
0.15
0.10
0
400
0
200
Pressure (hPa)
0
0
Pressure (hPa)
300
Fraction of points
0.60
0.40
0.30
0.20
0.15
0.10
200
p. GEOS5 20-year
0
100
Pressure (hPa)
1.00
Fraction of points
o. GEOS5 20-day
100
Fraction of points
n. GEOS5 2-day
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
Figure 5. (continued). Note that panel (a) is repeated for ease of comparison.
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 50
a. 2-day hindcasts
b. 20-day hindcasts
0.70
CZ
20
NA
0.65
CZ
CN
NA
E3
MR
MI
MO
r = 0.43
GI
0.60
0.55
0.50
CC
MOMR
GI
E3
CN
MI
14
r = 0.80
11
8
CC
0.45
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
Pattern correlation of net moistening (2-day hindcasts)
d. 20-day and 20-year pattern correlations
0.95
GI
MR
0.85
E3
CN
CZ
0.75
NA
0.65
r = 0.69
MI MO
0.55
5
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
CC
0.45
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
Pattern correlation of net moistening (20-year simulations)
Pattern correlation of net moistening (20-year simulations)
c. 20-year climate simulations
Fidelity in 20-year simulations
17
0.80
GI
0.70
E3
MR
0.60
CZ
r = 0.76
0.50
0.40
MO
MI
0.30
CC
NA
0.20
CN
0.10
0.00
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
Figure 6. For each GCM, pattern correlations of the net moistening diagnostics are computed
against ECMWF-YOTC (Fig. 5a). (a) shows the pattern correlation of net moistening from the
2-day hindcasts against the pattern correlation of 2-day lead time rainfall Hövmollers against
TRMM (as in Fig. 1b); (b) shows the pattern correlation of net moistening from the 2-day
hindcasts against MJO fidelity in the 20-day hindcasts; (c) shows the same as (b), but for the
20-year climate simulations; (d) compares the pattern correlations of net moistening from the
20-day hindcasts and 20-year climate simulations. Colored codes show relative fidelity in (b)
20-day hindcasts, (c) 20-year climate simulations and (d) 20-day hindcasts for the first letter and
20-year climate simulations for the second letter.
D R A F T
April 1, 2015, 9:40am
D R A F T
D R A F T
NA
GI
GEOS5
GISS-E2
20 min
30 min
0.625◦ ×0.5◦ , L72
2◦ ×2.5◦ , L40
NASAh
NASAh
60 min
30 min
45 min
T63 (1.9◦ ), L35
T127 (1.4◦ ), L31
T255 (0.7◦ ), L91
CC
CN
E3
CanCM4
CNRM-AM
ECEarth3
30 min
1.25◦ ×0.9◦ , L30
UCSDc,
LLNLd
CCCmae
CNRMf
SMHIg
CZ
Timestep
30 min
Resolution
1.25◦ ×0.9◦ , L30
Code Center
C5
NCARb
CAM5-ZM
with details.
Abbrev.
CAM5
Schmidt et al. [2014]
Rienecker et al. [2008]
Merryfield et al. [2013]
Voldoire et al. [2013]
Hazeleger et al. [2012]
Song et al. [2012]
Reference
Neale et al. [2012]
MetUM-GA3 MO
MIROC5
MI
MOHCi
0.83◦ ×0.56◦ , L70 12 min
Walters et al. [2011]
◦
AORI,
T85 (1.4 ), L40
12 min
Watanabe et al. [2010]
NIES,
JAMSTECj
k
MRI Atmospheric GCM
3
MRI-AGCM3 MR MRIk
TL159 (1.1◦ ), L48 30 min
Yukimoto et al. [2012]
a
CAM5 2-day hindcasts from CAM5 are combined with CAM5-ZM 20-day hindcasts and 20-year climate simulations.
b
National Center for Atmospheric Research
c
University of California, San Diego
d
Lawrence Livermore National Laboratory
e
Canadian Centre for Climate Modelling and Analysis
f
Centre National de Recherches Météorologiques
g
Swedish Meteorological and Hydrological Institute
h
National Aeronautics and Space Administration
i
Met Office Hadley Centre
j
Atmosphere–Ocean Research Institute (AORI), National Institute for Environmental Studies (NIES) and Japan Agency
for Marine Earth Technologies (JAMSTEC)
k
Meteorological Research Institute
points (L), timestep and a reference
Model name
Ver.
Community
Atmospheric 5
Modela
CAM5 with convective 5.1
microphysicsa
Canadian Coupled Model
4
f
CNRM Atmospheric Model 5.2
European
Community 3
Model
Goddard Earth Observing 5
System GCM
Goddard Institute for Space E2’
Studies GCM
Met Office Unified Model
GA3
Model for Interdisciplinary 5
Research on Climate
native horizontal resolution (longitude × latitude or wavenumber truncation (T) with equivalent in ◦ ) and number of vertical
Table 1. For each GCM: the full name, version, abbreviation used in the text, code used in the figures, contributing center,
April 1, 2015, 9:40am
X - 51
D R A F T
Table 2.
For each model, the pattern correlations of Hövmoller diagrams of daily-mean rainfall, averaged 10◦ S–10◦ N,
are computed against similarly constructed diagrams from TRMM. Correlations are computed from 20-day hindcast data,
using only start dates from the 2-day hindcast experiment, at 2-day and 10-day lead times. The 10-day lead time is the same
diagnostic used in Klingaman et al. [2015], except for only the 2-day hindcast start dates rather than all 20-day hindcast start
dates. Pattern correlations are computed over both 60–160◦ E and 60–105◦ E. For each set of pattern correlations, the rank of
with the ranks.
Classifications
20-day
20-year
Moderate Moderate
Moderate
Lower
Moderate Moderate
Moderate
Higher
Higher Moderate
Lower
Lower
Higher Moderate
Lower
Lower
Moderate
Higher
each model is shown for clarity. The classifications of each model in the 20-day hindcasts and 20-year climate simulations are
shown in the far-right columns, using the color-coded system described in the text, for comparison
Model
Day 2 (60◦ –160◦ E) Day 2 (60◦ –105◦ E) Day 10 (to 160◦ E) Day 10 (to 105◦ E)
Corr
Rank Corr
Rank Corr
Rank Corr
Rank
0.637
5/9 0.655
5/9 0.008
9/9 -0.193
9/9
0.615
7/9 0.600
8/9 0.256
5/9 0.107
6/9
4/9 0.176
7/9 0.093
7/9
0.656
3/9 0.671
0.674
1/9 0.684
2/9 0.280
3/9 0.168
4/9
0.653
4/9 0.705
1/9 0.266
4/9 0.173
3/9
0.633
6/9 0.633
6/9 0.187
6/9 0.127
5/9
0.659
2/9 0.681
3/9 0.396
1/9 0.190
2/9
0.496
9/9 0.492
9/9 0.104
8/9 0.064
8/9
0.593
8/9 0.601
7/9 0.395
2/9 0.349
1/9
CNRM-AM
MetUM-GA3
ECEarth3
MRI-AGCM3
CAM5-ZM
MIROC5
GEOS5
CanCM4
GISS-E2
D R A F T
April 1, 2015, 9:40am
D R A F T
X - 52

Vertical structure and physical processes of the MaddenâJulian

Transcription

Similar documents

S2S – MJOTF Joint Research Project White Paper MJO and

Your guide to managing your stock plan on NetBenefits

Stock Appreciation Rights

Know more about your equity plan than you ever could before.

eAdvisors Take the Lead - Fidelity Institutional Wealth Services

how to accept your grants/awards online

A new technology that has you protected Simple and

NWP, Climate Modeling, and Observational Studies at FSU With

Advantages and Disadvantages of Self-Employment

Vertical structure and physical processes of the MaddenâJulian

Transcription

Similar documents

Vertical structure and physical processes of the MaddenâJulian