Introduction

Download Report

Transcript Introduction

Dynamical Forecasting 2
Details for Today:
DATE:
BY:
27th January 2005
Mark Cresswell
FOLLOWED BY:
Nothing
69EG3137 – Impacts & Models of Climate Change
Lecture Topics
•
•
•
•
•
•
GARP
PROVOST
TOGA
GOALS
Impact of ENSO
DEMETER
• The future
GARP
Global
Atmospheric
Research
Programme
Set up by the World Meteorological Organization (WMO) to
examine the physical basis of climate and implications for
climate modelling – published document in 1975
Key section was by Edward Lorenz dealing with climate
predictability
GARP
Defined “climate” as an ensemble (collection) of all states
observed during some finite time period
Climate “prediction” must therefore be seen as the process of
determining how this ensemble will change at some point
As perfect measurements are impossible, the predictability of
any non-periodic system decays to zero as the range of
prediction becomes infinite
Unpredictability is caused by instability of small perturbations
that become subject to chaos
GARP
Small errors in the representation of initial conditions would
tend to double in amplitude every four days during a forecast
Eventually errors become no greater than guesswork – I.e.
randomly selecting an atmospheric state as a prediction
Perfect forecasting of large-scale atmospheric features would
require perfect representation and forecasting of smaller
scale features – impossible as models have too crude a
spatial scale to resolve such processes
GARP
The sea surface temperature (SST) is a key boundary
condition likely to provide most predictability
If the ensemble of weather patterns associated with one SST
pattern differs more than trivially with those associated with
another SST pattern then forecasts of positive skill should be
possible
A decrease (degradation) in spatial resolution of a GCM by a
factor of two can speed up the model by possibly a factor of
ten
PROVOST
PRedictability Of climate Variations On Seasonal to interannual Timescales
EC collaborative project – UKMO contribution was great
15-years (1979 to 1993) – sets of 4-month range, 9 member
ensemble integrations from the HadAM3 AGCM
Used prescribed ideal (observed) SST data – so simulations
are thus regarded as providing “potential” skill
PROVOST
Skill was found to be highest in the tropics
Skill in the extratropics was found to be in Spring (MAM)
Scores for precipitation are generally lower than temperature
ENSO forcing has a marked global impact on model
predictability
A substantial proportion of the skill achieved using observed
SSTs is retained using a persisted SST – suggesting that
persisted SST anomalies from the month preceding the start
of an integration could be viable for real-time forecasting
PROVOST
Prediction of model skill may be approached through the
relationship between ensemble spread and the skill of the
ensemble mean
The spread of members about the mean is a measure of the
sensitivity to initial conditions
Low ensemble spread is associated with high ensemble
mean skill
TOGA
Tropical Ocean and Global Atmosphere research programme
Evolved from loosely coordinated research efforts in the early
1970s – began as US effort in 1983 and became
international in 1985
Examined the mechanism for ENSO and the way wind stress
(and hence tropical Pacific SSTAs) was coupled to trade
wind strength
Key of TOGA was examination of tropical ocean –
atmosphere system predictability
GOALS
Global Ocean-Atmosphere-Land System
The TOGA programme was only partially successful and so
was replaced by GOALS in the mid-1990s
Aims:
•What are structure and dynamics of annual cycle of O-A-L
system and reasons for its variability over the globe?
•Relationship of variability to annual cycle
•Nature of tropical-extratropical interactions
•How might models be improved
Impact of ENSO
The El Niño Southern Oscillation (ENSO) may have a warm
phase (El Niño) or cool phase (La Nina). In both cases it
represents a warm or cold SST anomaly in the Eastern
Pacific.
Owen and Palmer (1986) produced the first empirical
evidence for the impact of ENSO on dynamical long-range
climate forecasts.
Based on two 3-member ensembles of 90-day forecasts for
1982-83 with the UKMO 11-layer GCM
Impact of ENSO
The first ensemble used observed SSTs and the second
used climatological SSTs.
ENSO SST anomalies can force a realistic and statistically
significant time-averaged response around the entire tropical
region
In the tropics, the skill was improved with observed SSTs for
all timescales. In the extratropics, skill was improved only on
a 30-day timescale
DEMETER
The European Centre for Medium Range Weather
Forecasting (ECMWF) near Reading has been leading the
way in new long-range prediction research
ECMWF houses a Fujitsu VPP5000 supercomputer with a
massively parallel 100-processor array.
Computer allows HOPE ocean model and EPS GCM to work
together as a completely coupled ocean-atmosphere climate
model. A new forecasting system is currently being
developed called DEMETER
DEMETER
Development of a European Multimodel Ensemble system
for seasonal to inTERannual prediction
The project aims to develop a well-validated European
coupled multi-model ensemble forecast system for reliable
seasonal to interannual prediction
Funded by the EU it brings together 6 european climate
modelling groups: ECMWF, Météo-France, LODYC, UKMO,
MPI, CERFACS, INGV and INM-HIRLAM. Each model is
installed and run at ECMWF.
Analysis and formulation is based on evidence from
PROVOST
DEMETER
Initialisation and validation is performed with the new ERA40 reanalysis dataset (1957 to 2001) that replaces the older
ERA-15 data - with its associated flaws.
By combining the 6 sets of ensembles (one from each
European model) it is hoped that simulations with greater
skill will be achieved.
Archives of forecast variables (including: air temperature, U
and V velocities, specific humidity, snow depth, cloud cover,
precipitation etc) will be made available for further research
and model verification
The Future
Following on from DEMETER, a new project started in
2003/2004 funded by the European Framework 6
University of Reading has begun to work on a new model
formulation for global climate model resolution of 40km.
Once installed at the Hadley Centre, this new model will be
used to examine future global climate change scenarios (up
to 100 years ahead) for specific regions
Improved assimilation will be possible with new satellite
remote sensing systems being planned and now on line (e.g.
METEOSAT Second Generation and GPM project)