HR: 15:30h
AN: A44A-01 [Abstracts]
TI: Diagnosing Sources for the Contiguous US Seasonal Forecast Skill
AU: * Quan, X
EM: quan.xiao-wei@noaa.gov
AF: NOAA-CIRES Climate Diagnostics Center, R/CDC1, 325 Broadway, Boulder, Boulder, CO 80305
AU: Hoerling, M P
EM: martin.hoerling@noaa.gov
AF: NOAA-CIRES Climate Diagnostics Center, R/CDC1, 325 Broadway, Boulder, Boulder, CO 80305
AU: Whitaker, J S
EM: Jeffrey.S.Whitaker@noaa.gov
AF: NOAA-CIRES Climate Diagnostics Center, R/CDC1, 325 Broadway, Boulder, Boulder, CO 80305
AU: Xu, T
EM: taiyi.xu@noaa.gov
AF: NOAA-CIRES Climate Diagnostics Center, R/CDC1, 325 Broadway, Boulder, Boulder, CO 80305
AB:
The skill in U.S. seasonal forecasts is widely held to arise first and foremost from the atmosphere's sensitivity to
anomalous sea surface temperatures. Much progress has been made in recent years in predicting the latter, especially at
one-season leads. General circulation models used in modern prediction methodologies also exhibit superior fidelity relative
to earlier generations. Notwithstanding these advances, a situation of modest U.S. seasonal forecast skill remains. To
interpret the current state of U.S. skill, our study diagnoses the nature of the SST-related skill and its origins.
The simulation skill of a multiple model, large ensemble suite of atmospheric GCM's using the monthly observed specified
global SSTs is first presented. Skill is stratified by season, region, and for the variables surface temperature (T) and
precipitation (P).
The skill sources are diagnosed by constructing a multi-variant CCA whose co-variance matrix relates tropical SST
"predictors' to the GCM simulated US T and P "predictands". This diagnostic model confirms that the bulk of skill is of
tropical SST origin, and is furthermore linearly related to those SSTs. The CCA model is subjected to various truncations,
from which it is discovered that the bulk of skill originates from a single ENSO pattern of SST predictor.
A second, additional source of US skill is discovered. This source is largely non-ENSO. The predictor consists of SST
variations in the subtropical west Pacific including the South China Sea, and explains much of the GCM's Fall seasonal
temperature skill in northern US.
A seasonal hindcast model is then constructed based on lagged CCA. Hindcast skill for 1950-99 reveals the majority of
simulation skill of the same period is realized using the above-mentioned two primary tropical SST predictors.
DE: 3309 Climatology (1620)
DE: 3319 General circulation
DE: 3339 Ocean/atmosphere interactions (0312, 4504)
DE: 3354 Precipitation (1854)
DE: 4215 Climate and interannual variability (3309)
SC: Atmospheric Sciences [A]
MN: 2005 Joint Assembly