Temperature, Salinity and Sea Level: climate variability from ocean reanalyses Intercomparison Items

1
Temperature, Salinity and Sea Levelclimate
variability from ocean reanalyses(Intercomparison
Items 3 4)
Magdalena A. Balmaseda Anthony Weaver
2
Outline

• Defining the variability signal and noise
• Seasonal cycle removed Anomalies with respect to
  the common period (1994-2000)
• 12-month/3-month running mean
• Focus on the upper ocean (upper 300m) Equator,
  Mid latitudes
• Temperature and salinity
• What can we say about climate variability?
• Time variation of uncertainty
• Outliers?
• Source of uncertainty (forcing, model,
  assimilation method)?
• Sea Level variations volume or mass changes?
• Summary and conclusions

3
Focus Regions for Items 3 4
4
Uncertainty in the Mean

• Ambiguity in the definition
• closest level, interpolated values?
• Real Uncertainty?

We will use anomalies
5
T300 Equatorial regions

• Eq Pac Uncertainty decreases with time.
• Relatively robust interannual variability.
• Increased uncertainty after 2000. Why?

6
What can we say about Tropical variability?
How should we interpret the outliers? Need to
understand the reasons An outlier for the good
reason is very valuable An outlier for the wrong
reason is damaging
7
T300 Mid latitudes (northern)

• The North Atlantic is dominated by a warming
  trend, especially post 1997
• Large uncertainty after 2000.
• Phase/amplitude of decadal variability is poorly
  resolved.

8
GLOBAL SIGNALS
Warming occurs in first 750m
9
Salinity Variability

• Salinity variability dominated by spin-up effects
  in several cases.
• There is no consistency among reanalyes

10
Signal/Noise Ratio (long period)
EQPAC EQATL EQINDTRPAC TRATL NPAC NATL GLOBAL
11
Signal/Noise Ratio (Short period, interannual)
EQPAC EQATL EQINDTRPAC TRATL NPAC NATL GLOBAL
12
Sources of Uncertainty
All ERA40 ERA40-No data ERA40-Assim
13
Equatorial Atlantic Assimilation increases spread
All ERA40 ERA40-No data ERA40-Assim
14
Assimilation and Uncertainty T300
1956-2005
EQPAC EQATL EQIND TRPAC

• Long period 1956-2005
• Assimilation decreases uncertainty everywhere

1993-2005
EQPAC EQATL EQIND TRPAC
TRATL NPAC NATL GLOBAL
Latest period 1993-2005 Only in EQPAC the
assimilation reduces uncertainty
TRATL NPAC NATL GLOBAL
15
Assimilation and Uncertainty S300
1956-2005
EQPAC EQATL EQIND TRPAC

• Long period 1956-2005
• Assimilation increases uncertainty everywhere

1993-2005
EQPAC EQATL EQIND TRPAC
TRATL NPAC NATL GLOBAL
Latest period 1993-2005 Assimilation increases
uncertainty
TRATL NPAC NATL GLOBAL
16
GLOBAL SEA LEVEL VARIATIONS I

• Prior to 2002, changes in sea level and steric
  height are correlated
• Volume rather than mass
• Changes in bottom pressure are not consistent
  amont analyses?

17
GLOBAL SEA LEVEL VARIATIONS II
Temperature contribution to dynamic height
dominates the trend
18
Sea level variations Seasonal cycle
Seasonal variations in global sea level are
attributed to mass variations that affect the
bottom pressure.
19
Summary and Conclusions

• There is large uncertainty in climate signals
• Signal to noise ratio gt 1 in the Eastern Pacific
  for Temperature
• Signal to noise ratio lt1 for salinity in most
  regions
• Warming trend in the 90s is consistently
  reproduced
• What is happening now? There is not consistent
  picture
• Forcing fluxes and analysis methods are largest
  source of uncertainty
• Data Assimilation does not always collapse the
  spread We need to pay more attention to the
  assimilation methods.
• Global Sea level trends
• 1993-2002 Consistently attributed to temperature
  rise.
• 2002-present It is not clear. Mass increase?
• Seasonal variations related to mass variations.

20
NEXT

• Web page to collect. WHERE? GSOP web page?
• Maps of signal to noise ratio, and so on
• Need to define the variables so there are not
  so dependent on the vertical model grid.
• There are currently 20 analysis are there more
  there?
• Agreement on colour coding and naming convention?
• Other?