Andrew W' Robertson

1
Seasonal Climate Prediction over the Americas

• Andrew W. Robertson
• International Research Institute for Climate
  Prediction (IRI)

2
Outline

• Observed precipitation relationships with SST and
  ENSO
• Two-tier seasonal prediction
• SST prediction coupled models
• Atmospheric response to SST forcing atmospheric
  GCMs
• Other issues
• Atlantic SST anomalies
• Antecedent land conditions
• Downscaling in space and time

3
.
Simultaneous correlations between seasonal
mean rainfall in the indicated region (yellow
dot) and SST anomalies
4
Simultaneous correlations between seasonal mean
rainfall in the indicated region (yellow dot)
and SST anomalies
5
Statistical streamflow prediction based on SST
Probabilistic Forecasts of 1993-2000 Jan-Dec
Annual Inflow into Oros from the preceding July.
The 1914-1991 data was used for model fitting.
De Souza and Lau (2003)
6
Probabilistic categorical forecasts
7
ENSO-assoc. Precip. probabilities Jan-Mar
La Niña
El Niño
8
ENSO-assoc. Precip. probabilities OctDec
Precip Oct-Nov-Dec
El Niño
La Niña
9
Pacific-South American Pattern
200-hPa Winds Seasons defined by ENSO influence
on Uruguay precip.
10
(No Transcript)
11
Niño-3 SST prediction skill
12
(No Transcript)
13
Multiple AGCM ensemble precipitation skill
195395
14
IRI Operational skill 19972001
OND
JFM
JAS
AMJ
1-month lead
15
AGCM precipitation weights Jul-Aug-Sep
16
ECHAM4 precipitation
17
GPCP precipitation
18
GPCP JAS precipitation
19
GCM JAS precipitation
20
Loss of skill due to imperfect SST prediction
L. Goddard, IRI
21
Forecast development

• Atlantic SST anomalies
• Antecedent land conditions
• Downscaling in space and time

22
South Atlantic SSTAs
Leading atmospheric EOF (Jan-Feb-Mar)
Regression with SST
23
AGCM Response to S. Atlantic SST Anomaly
24
Near-decadal NAO influence
N.-Paegle, Robertson Mechoso (2000)
25
Antecedent land conditions
New Mexico JulAug precip. vs. Spring snow pack
(SWE) (Gutzler 2000)
195160 199195
196190
26
Regional downscaling of precipitation over NE
Brazil
RSM simulations over South America, forced by
ECHAM4 Courtesy of L. Sun, IRI
27
Frequency of dry-spell lengths
RSM simulations over South America forced by
ECHAM4 Courtesy of L. Sun, IRI
28
Precipitation extremes
Predictability of 90th percentile using Pacific
SSTs (Gershunov Cayan 2003)
29
Outstanding problems for prediction

• Forecasts of ENSO SSTAs over the tropical Pacific
• Coupled modeling is crucial (CGCMs, intermediate
  coupled models)
• SST prediction outside the tropical Pacific
• Slab ocean models may be sufficient
• Atmospheric model performance, esp. convection
  over land
• physical parameterizations
• Antecedent land conditions