Seasonal Climate Prediction

1
Seasonal Climate Prediction

• Youmin Tang
• Environmental Science and Engineering, University
  of Northern British Columbia

2
Contents

• Introduction
• Basic theory and methods for the dynamical
  climate prediction system
• International activities on the seasonal climate
  prediction
• Seasonal prediction in Canada
• Climate prediction in UNBC
• Future Challenges

3
Three types of predictions

• Weather Prediction
• Prediction of long-term climate change
• Seasonal Prediction

4
Prognostic diagnostic Eq.
u, v, T, S
w, p, ?
5
(No Transcript)
6
Differences

• NWP time evolution of the exact state of the
  atmosphere
• Long-term prediction gross features of a changed
  climate averaged over many years
• Seasonal Predictions describe statistic aspects
  of atmospheric anomalies over 1-3 months

7
Weather prediction

• Goal to forecast the exact state of the
  atmosphere from initial conditions, at high time
  resolution over several days.
• Combination of statistical technique, experience,
  and intuition, in the final stage of most
  forecast.
• Mid-latitude, the limit predictability is usually
  considered to be 10-14 days (Lorenz, 1982).

8
Weather prediction

• Predictability arises solely from internal
  atmospheric dynamics
• Accurate atmospheric initial conditions
• Smaller errors in the initial state can grow
  rapidly and lead to a poor forecast even with a
  perfect model
• Slightly different initial conditions are used
  for ensemble forecast
• Slowly evolving lower boundary conditions are
  often assumed to be constant

9
Prediction of long-term climate change

• Goal to characterize changes in the long-term
  mean atmospheric and oceanic circulation and
  especially to characterize mean changes at the
  earths surface
• Tools a coupled atmosphere-ocean-land-ice model
  --- Climate system model
• Concern with gross features of a changed climate
  averaged over many years

10
Seasonal Prediction

• Focus fairly qualitatively on a few key climate
  variables
• Surface temperature
• Precipitation
• Distinct from both NWP and Climate Simulation in
  three aspects
• In Purpose
• In Approaches
• In timescale

11
The New Challenge Linking Climate to Weather
12
Why Seasonal Prediction
Growing demand for reliable seasonal forecasts
energy
13
Venezuela
Germany
India
14
(No Transcript)
15
(No Transcript)
16
Use of better Drought forecasts to Improved Dam
and hydropower management for hydroelectric power
generation has enabled the domestic and
manufacturing industries in Kenya and Tanzania to
run at optimum capacity
17
Economic Loss caused by 1982/1983 El Nino Event
18
Predictability of seasonal prediction
Obs.
Internal chaotic Process
Theoretical limit
Method (model) Quality of IC BC
Potential predictability
Actual predictability
In climate prediction, Potential predictability
is usually regarded as the predictability with
full information of future boundary condition
(e.g., SST). Thus, predictability is varied with
similarity between the response of real
atmosphere and prediction method to the same BC.
From Prof. In-Sik kang
19
Why --- limit of Predictability

• The limitation of predictability arises from
• The imperfections of the forecast model
• The nonlinearity of the climate system
• The predictive skill depends on
• The field considered
• The model used for forecast
• The initial state of the system

20
Limitation of Predictability

• Any statement about the predictive skill should
  include the word for this model
• Predictive skill should be calculated for a
  large number of situations in order to make the
  most general statement possible

21
Methods for seasonal prediction

• Statistical Method
• Dynamical Method
• Hybrid Method

22
Statistical Methods for seasonal prediction

• Since the beginning of the twentieth century
  (e.g., Quayle, 1929)
• Based on historical data and employ a
  mathematical relationship between predicted and
  predictor variables.
• SST anomalies (especially over the tropical
  Pacific Ocean) are the sole predictor for the
  statistical forecasts of seasonal climate
  anomalies (e.g., Folland et al., 1991 Ward and
  Folland, 1991 Barnston, 1994)

23
Statistical Methods for seasonal prediction

• Regression approaches
• e.g., Knaff and Landsea, 1997
• Canonical correlation analysis (CCA)
• e.g., Barnston and Ropelewski, 1992
• Neural network models
• e.g., Sahai et al., 2000 Tanggang et al, 1998,
  Tang et al. 2001 2002 2003.

24
Statistical Methods for seasonal prediction

• Limitations
• require a long and accurate data of the earths
  climate
• require an understanding of the physically based
  relationships between predicted and predictor
  variables
• Unstable relationship (e.g. Krishna Kumar, 1999 )
• No physics

25
Dynamical Methods for seasonal prediction

• Since late 80s last century
• Based on mathematical representation of physical
  laws governing the behavior of the atmosphere or
  the coupled atmosphere-ocean system.

• Be able to estimate uncertainty of prediction
• through Ensemble prediction.
• High potential to improve skill in the future

26
Seasonal-to-interannual prediction

• The Basis for all predictions at timescales
  longer than a month is the hypothesis that, on
  these timescales, the atmospheric statistics are
  in equilibrium with the surface boundary
  conditions
• A prediction of Surface Boundary conditions will
  lead to some statistical knowledge of the
  atmosphere

27
Seasonal-to-interannual prediction

• Slowly varying boundary conditions, impose a slow
  variation of atmospheric statistics
• Sea surface temperature
• Soil moisture
• Sea ice extent
• Surface albedo

28
Boundary conditions

• Strong interact with the atmosphere
• Soil moisture --- rainfall evaporation
• Albedo --- Snow and ice extent
• SST --- fluxes of heat and momentum from
  atmosphere
• Evolve with their own dynamics
• A climate system model is needed

29
(No Transcript)
30
(No Transcript)
31
(No Transcript)
32
(No Transcript)
33
Operational prediction-Two tiers
34
Ensemble Prediction

• The Ensemble prediction simulates possible
  initial uncertainties by adding, to the original
  analysis, small perturbations within the limits
  of uncertainty of the analysis. From these
  alternative analyses, a number of alternative
  forecasts are produced

35
Ensemble runs? How to optimally perturb system?

• The model dimensionality is large, typically
  106.
• We must perturb the system wisely such that we
  can use affordable perturbation members for
  ensemble predictions.
• We need to find the optimal perturbation
  patterns? singular vectors or breeding vectors of
  the linearized operator of the original system.

36
(No Transcript)
37

• The growth (forecast error) of perturbation (or
  initial error) in the time interval can be
  expressed in the form

38
(No Transcript)
39

• The vector of the small perturbation
• that maximizes is the
• first eigenvector of ,i.e, the
  singular vector of A. where A is the ad joint
  operator of A.

40
International Research Activities on the
dynamical Seasonal prediction
41
Prediction and predictability
42
International Projects

• CLIVAR (Study of CLImate VARiability and
  Predictability)
• SMIP ( Seasonal Prediction Model Intercomparison
  Project --- Phase I Phase II )
• NSIPP (NASA Seasonal to Interannual Prediction
  Project)
• PROVOST (PRediction Of Climate Variation On
  Seasonal to interannual Time scale)
• DEMETER(Development of a European Multimodel
  Ensemble system for Seasonal to interannual
  prediction)
• APCN Multi-model Ensemble Project

43
Development of a European Multi-Model Ensemble
System for Seasonal to Interannual
Prediction (http//www.ecmwf.int/research/demeter
/general/index.html)
44
Multi-Model Ensemble System

• DEMETER system 6 coupled global circulation
  models

9 member ensembles ERA-40 initial conditions
SST and wind perturbations 4 start dates per
year 6 months hindcasts
Hindcast production for 1987-1998 (1958-2001)
45
APCN Multi-Model Ensemble System
46
Participating Models
Member Economies Acronym Organization Model Resolution
Australia POAMA Bureau of Meteorology Research Centre  T47L17
Canada MSC Meteorological Service of Canada 1.875 ? ? 1.875 ? L50
China NCC National Climate Center/CMA T63L16
China IAP Institute of Atmospheric Physics 4 ? ? 5 ? L2
Chinese Taipei CWB Central Weather Bureau T42L18
Japan JMA Japan Meteorological Agency T63L40
Korea GDAPS/KMA Korea Meteorological Administration T106L21
Korea GCPS/KMA Korea Meteorological Administration T63L21
Korea METRI/KMA Meteorological Research Institute 4 ? ? 5 ? L17
Russia MGO Main Geophysical Observatory T42L14
Russia HMC Hydrometeorological Centre of Russia 1.125 ? ? 1.40625 ? L28
USA COLA Center for Ocean-Land-Atmosphere Studies T63L18
USA IRI International Research Institute for Climate Prediction T42L18
USA NCEP NCEP Coupled Forecast System T62L64
USA NSIPP/NASA National Aeronautics and Space Administration 2 ? ? 2.5 ? L34
47
Research Institutions

• International Research Institute for Climate
  Prediction (IRI) (Initiated in 1994)
• Climate Prediction Center(CPC/NCEP)
• ECMWF (Initiated in 1995)
• UK Met office (Initiated in 1987, Ward and
  Folland, 1991))
• CCCma (Canada)
• RPN (Canada)
• BMRC (Australia)
• Experimental climate prediction center(ECPC),
  Scripps Institute of Oceanography
• Korea Meteorology Administration (KMA)
• Japan Meteorology Administration (JMA)

48
Seasonal Prediction in Canada

• Since September 1995, the Canadian Meteorological
  Centre has been producing 0-3 month outlooks for
  Canada.
• The seasonal forecast results from an ensemble of
  12 model runs 6 runs from a Global Environmental
  Multiscale model (GEM) of RPN, that has a
  horizontal resolution of 1.875 degrees with 50
  vertical levels, and 6 runs from a Climate model
  (GCM2) of the CCCma.

49
Surface Air Temperature Forecast

• an average of the daily temperature as predicted
  by the models.
• The climatologies of the models are then
  subtracted from the mean forecast seasonal
  temperatures to derived the forecast anomalies of
  each model. The anomalies of the two models are
  then normalized and combined using an arithmetic
  average.
• The anomalies are divided in three categories
  (above, near and below the normal).

50
Precipitation Forecast

• The forecasts are made using the total
  accumulated water precipitation over the season.
  The precipitation predicted is the total liquid
  and includes all types snow, rain, ice pellets,
  etc. The climatology of the models is subtracted
  from the total precipitation forecast to derive
  the anomalies. The anomalies of the two models
  are then combined using a simple normalized
  average. Finally the precipitation anomalies are
  divided in three categories (above, near and
  below the normal) as is done for the temperature
  anomaly forecast.

51
Skill of Summer T
52
Skill of Winter T
53
Skill of Summer P
54
Skill of Winter P
55
The prediction is issued operationally

• http//meteo.ec.gc.ca/saisons/index_e.htmlclimato
  logy

56
Prediction and Predictability of the Global
Atmosphere-Ocean Systemfrom Days to Decades

• A Five-Year Network in Canada funded by
  Canadian foundation of Climate and Atmospheric
  Sciences.

57
(No Transcript)
58
Climate Prediction and Predictability in UNBC

• Dr. Youmin Tang (group leader)
• Dr. Ziwang Deng
• Dr. Xiaobing Zhou
• Peter Mills
• Jasion Ambadan
• YangJie Cheng
• Zhiyu Wang

59
Ocean model

• OPA8.1 OGCM
• 25 layers in the vertical direction with 17
  concentrated in the top 250m of the ocean.
• Model domain 30N - 30S and 120E - 75W.
• Resolution 1 degree in the zonal direction in
  the meridional direction, the resolution is 0.5
  degree within 5 degree of the equator, smoothly
  increasing up to 2.0 degree at 30N and 30S.
• Time step is 1.5 hour.

60
Atmospheric models

• Model1 Statistical model. ? HCM1
• Model2 Intermediate complexity dynamical
• model ? HCM2
• Tang et al. 2004, J. Geophy. Res (ocean),
  109, C05014)
• Tang et al. 2004, Geophy. Res. Letters, Vol.
  30, No. 13, 1694
• Tang et al. 2004, J Phys. Oceangr. Vol 34,
  No. 3, 623-642.

61
(No Transcript)
62
(No Transcript)
63
(No Transcript)
64
(No Transcript)
65
(No Transcript)
66
(No Transcript)
67
(No Transcript)
68
Future Challenges

• Extending the geographical range of SST
  predictions
• Model initialization
• Land data assimilation system
• Oceanic data assimilation system
• Ensemble runs
• Multi-model ensemble technique
• Model development
• Progress in dynamical seasonal prediction in the
  future depends critically on improvement of
  coupled ocean-atmosphere-land models

69
(No Transcript)
70
Krishnamurti et al.(2000)
71
Multi-model hindcasts
ACC for Niño3 SST for Multimodel, ECMWF, MetFr,
MetOf, and LODYC