Predictability of Week-2 to Monthly Climate

1
Predictability of Week-2 to Monthly Climate
NOAA Earth System Research Laboratory

• Tom Hamill ( Jeff Whitaker, Randy Dole)
• NOAA Earth System Research Lab, Physical Sciences
  Division (Formerly Climate Diagnostics Center)
• tom.hamill_at_noaa.gov

2
Disclaimer
Statistics and Ensembles
Climate Dynamics
I am no 44.45 kg (98-pound) weakling!
I know nothing!
3
Topics

• Main sources of potential predictability
• Tools for improving 2-4 week predictions, with a
  focus on reforecasts

4
Sources of potential predictabilitythe
atmospheric initial condition
Still some skill in week 2. Will continued
NWP improvements affect weeks 3 and 4 positively?
from LaLaurette et al (2006), ECMWF tech memo
5
Views of predictability Inverse error cascade
or baroclinic error growth?
Growth of errors in Lorenz (1969) experiment
assuming -5/3 power law for spectrum of
energy per unit wavenumber. Implies very rapid
upscale cascade of errors, fully saturating by
week 2. However, -3 power law more typical for
synoptic scales.
6
Views of predictability inverse cascade or
baroclinic error growth?
Tribbia and Baumhefner (2004) Rapid error
growth between 0 - 0.25 days. Thereafter, more
exponential til saturation. In our alternative
view the inverse cascade becomes of lesser
importance than in the traditional view. The
primary role of the inverse cascade is to seed
disturbances in the baroclinically active region
of the spectrum. Once in the synoptic scales, the
errors organize within the synoptic structures
and amplify, extracting energy from the
large-scale background flow, not the smaller
scales.
7
Sometimes predicting planetary scales well will
dramatically affect skill
Positions of fronts every 24h
unblocked
blocked
from Pettersen 1956
8
Sources of predictabilitythe stratospheric
initial condition
Baldwin and Dunkerton (1999)
9
NAO surface pattern predicted by previous months
10 hPa AO?
But (1) convincing dynamical explanation of this
not yet proposed, and (2) presumably the
mechanism involves large-scale dynamics, so they
wouldnt well-represented in current models?
Hence will this be a future source of improved
predictability?
Baldwin and Dunkerton (1999)
10
Sources of predictabilityENSO
11
How tropical heating can affect extratropical
circulations
Schematic of the dominant changes in the upper
troposphere in response to increases in SSTs,
enhanced convection, and anomalous upper
tropospheric divergence in the vicinity of the
equator (scalloped region). Anomalous outflow
into each hemisphere results in subtropical
convergence and an anomalous anticyclone pair
straddling the equator, as indicated by the
streamlines. A wave train of alternating high
and low geopotential and streamfunction anomalies
results from the quasi-stationary Rossby wave
response (linked by the double line). In turn,
this typically produces a southward shift in the
storm track associated with the subtropical jet
stream, leading to enhanced storm track activity
to the south (dark stipple) and diminished
activity to the north (light stipple) of the
first cyclonic center.
from Trenberth et al (1998) see also Sardeshmukh
and Hoskins (1988)
12
Altered position of jet in El Niño vs. La Niña
El Niño
La Niña
from Shapiro et al. (2001)
13
Characteristics of cyclones changes from to -
phase of ENSO
300 K PV (colors) on 6 Feb 1998
1998 El Niño
forward-breaking PV wave
1999 La Niña
rearward-breaking PV wave
from Shapiro et al. (2001)
14
El Niño and hurricane activity
inactive
El Niño is one of many large-scale predictors of
western-Atlantic tropical cyclone
activity. ENSO forces increased vertical wind
shear in trades.
active
Blake and Gray (2004)
15
July patterns for active August-October hurricane
season in U.S.
July 925 - 400 hPa average wind pattern in the
month preceding
August wind pattern associated with the
top quintile of years with U.S. hurricane
landfalls
35-40 skill relative to climatology
Northward displacement of Bermuda high, Rockies
anticyclone, La Niña signature, etc..
Saunders and Lea, 2005
16
Sources of potential predictability
Madden-Julian Oscillation
orange-yellow are cold cold tops
17
MJO the optimistic perspective (Ferranti et al.
1990)
relax tropics to analyzed state control relax
tropics to persisted initial condition
ECMWF T42 forecasts in times of active MJO test
skill of perfect tropical forcing vs. existing
model. However (1) small sample size, (2) were
not yet anywhere near perfect tropical
prediction, (3) MJO active only small percentage
of time.
18
91-day running mean of RMM12 RMM22
long periods with no significant MJO activity
?
from Wheelers http//www.bom.gov.au/bmrc/clfor/cf
staff/matw/maproom/RMM/ts.PCvar91drm.gif
19
Predictability from land-state anomalies?
Case 2004 European heat wave (dry soils
preceding that spring)
1-2 month response to observed SST forcing
(warmer than avg).
response to dry initial root layer (note EMCWF
model diminishes amplitude of seasonal soil
moisture perturbations, so had to force dry soils
in model)
response to dry soil over all layers
Ferranti and Viterbo (2006)
20
Tools for improving weeks 3-4 predictions

• Linear Inverse and other statistical models (see
  Newman, Penland, Sardeskmukhs work).
• Reforecasts (Ill concentrate on this).

21
Linear Inverse Modeling

• Good for diagnosing
• sources of linear
• predictability.
• Better than circa 1998
• MRF (GFS) in week 3.
• But models may improve,
• LIM wont much.

Newman et al. (2003)
22
Tools for improving predictionsreforecast data
sets (e.g., our prototype with the T62 model)

• Reforecast definition a data set of
  retrospective numerical forecasts using the same
  model as is used to generate real-time forecasts.
• Model T62L28 NCEP GFS, circa 1998
• Initial States NCEP-NCAR Reanalysis II plus 7
  /- bred modes.
• Duration 15 days runs every day at 00Z from
  19781101 to now. (http//www.cdc.noaa.gov/people/j
  effrey.s.whitaker/refcst/week2).
• Data Selected fields (winds, hgt, temp on 5
  press levels, precip, t2m, u10m, v10m, pwat,
  prmsl, rh700, heating). NCEP/NCAR reanalysis
  verifying fields included (Web form to download
  at http//www.cdc.noaa.gov/reforecast).

23
WHY?
850 hPa temperature bias for a grid point in
the central U.S.
Spread of yearly bias estimates from
31-day running mean F - O Note the spread is
often larger than the bias, especially for long
leads.
24
Calibrating ensemble forecasts raw, bias
correction, logistic regression
Hamill et al. (2004)
25
surface temperature
precipitation
26
Calibrating Z500 CRPSS
27
Calibrating T850 CRPSS
28
Calibrating T2m CRPSS
29
Are reforecasts still necessary with improved
models?
ECMWF produced a short reforecast data
set. Calibration using their week-2 reforecasts p
roduced a skill increase of 11 for our
reforecast, skill improvement was 16
Whitaker and Vitart (2006)
30
Multi-model reforecasts
may be worth keeping old reforecasts around
Whitaker and Vitart (2006)
31
Reforecasts as a tool for studying
predictability
Hamill et al. (2006)
32
Skill of predicting these patterns
Hamill et al. (2006)
33
General areas where more research is needed to
improve weeks 3-4 forecasts.

• What is the value of coupled ocean / atmosphere
  models for these leads?
• Are the predictable aspects in weeks 3-4 that
  emerge from nonlinear ensemble forecasts ones
  that are essentially linear and can be obtained
  by cheaper LIM techniques?
• What are the mechanisms for error growth of
  planetary scales?

34
My hunches

• Still some sources of predictability that are
  nonlinear in weeks 3-4, hence NWP will be useful
  supplement to purely statistical techniques. But
  we should be systematic in determining whether
  nonlinear effects are important.
• However, if youre adopting an NWP approach,
  reanalyses / reforecasts will be crucial. Model
  biases are huge at these leads.

35
References
Hamill, T. M., J. S. Whitaker, and X. Wei, 2003
Ensemble re-forecasting improving medium-range
forecast skill using retrospective forecasts.
Mon. Wea. Rev., 132, 1434-1447.
http//www.cdc.noaa.gov/people/tom.hamill/reforeca
st_mwr.pdf Hamill, T. M., J. S. Whitaker, and
S. L. Mullen, 2005 Reforecasts, an important
dataset for improving weather predictions. Bull.
Amer. Meteor. Soc., 87, 33-46. http//www.cdc.noaa
.gov/people/tom.hamill/refcst_bams.pdf
Whitaker, J. S, F. Vitart, and X. Wei, 2006
Improving week two forecasts with multi-model
re-forecast ensembles. Mon. Wea. Rev., 134,
2279-2284. http//www.cdc.noaa.gov/people/jeffrey.
s.whitaker/Manuscripts/multimodel.pdf Hamill,
T. M., and J. S. Whitaker, 2006 Probabilistic
quantitative precipitation forecasts based on
reforecast analogs theory and application. Mon.
Wea. Rev., in press. http//www.cdc.noaa.gov/peopl
e/tom.hamill/reforecast_analog_v2.pdf Hamill,
T. M., and J. Juras, 2006 Measuring forecast
skill is it real skill or is it the varying
climatology? Quart. J. Royal Meteor. Soc., in
press. http//www.cdc.noaa.gov/people/tom.hamill/s
kill_overforecast_QJ_v2.pdf Wilks, D. S., and
T. M. Hamill, 2006 Comparison of ensemble-MOS
methods using GFS reforecasts. Mon. Wea. Rev., in
press. http//www.cdc.noaa.gov/people/tom.hamill/W
ilksHamill_emos.pdf Hamill, T. M. and J. S.
Whitaker, 2006 White Paper. Producing
high-skill probabilistic forecasts
using reforecasts implementing the National
Research Council vision. Available at
http//www.cdc.noaa.gov/people/tom.hamill/whitepap
er_reforecast.pdf .
36
1st EOF (RMM1) and 2nd EOF (RMM2) of MJO
from Wheelers http//www.bom.gov.au/bmrc/clfor/cf
staff/matw/maproom/RMM/eof1and2.htm
We performed EOF analysis on the combined daily
fields of equatorially-averaged (15S to 15N)
OLR, 850hPa zonal wind, and 200 hPa zonal wind
for the period of 1979 to 2001 (23 years). We did
this on the covariance matrix with each field
normalized by the square-root of its global mean
variance first. This is necessary so that each
field contributes the same amount of variance to
the combined field. Before the EOF analysis,
however, we also performed the following
Remove the long-term mean and climatological
seasonal cycle (3 harmonics) from each field at
each grid-point. Remove the variability
associated with El Nino (that which is linearly
related the ENSO SST1 index). Remove a
120-day mean of the most recent 120 days at each
point..
back
37
Predictability from tropical boundary conditions
(SSTs)

• Rossby wave energy dispersion from steady,
  localized heat sources

Sardeshmukh and Hoskins (1988) (a) divergence
perturbation and (d) streamfunction perturbation
on day 48 assuming fully nonlinear response
Horel and Wallaces (1981) schematic illustration
of geopotential height anomalies introduced by
warm SSTs along the dateline.
38
MJO a note of caution
39
MJO a note of caution
40
MJO a note of caution
41
Issues (1) should reanalyses be part of
reforecast process?

• Want homogeneous
• characteristics of forecasts skill the same for
  1980s forecasts as 2006 forecasts.
• Part of better skill of current forecasts is the
  better initial condition.
• Reanalysis would improve skill of old forecasts.
• Reanalyses should use same or similar model as
  used in reforecasts.