Sea Surface Temperature SST Analyses for Climate

1
Sea Surface Temperature (SST) Analyses for Climate

• Richard W. Reynolds
• National Oceanic and Atmospheric Administration
• National Environmental Satellite, Data
  Information Service
• National Climatic Data Center
• Asheville, NC
• E-mail Richard.W.Reynolds_at_noaa.gov

National Climatic Data Center
2
Introduction

• Why should we care about SST?
• What are the types of SST data?
• In situ (measured from ships and buoys)
• Satellite
• How are the data distributed?
• How are SST analyses computed?
• What are the SST errors?

National Climatic Data Center
3
Climate Change Detection

• Surface Temperature
• Land plus SST
• Marine air temperatures not used because biased
  noisier than SST

Smoothed annual anomalies of global combined
land-surface air and sea surface temperatures
(oC).
Vertical axis temperature anomaly Horizontal
axis time
4
SSTs for El Niño and La Niña
Total SST oC
Anomaly SST Anomaly is departure from normal
5
El Niño vs. Normal

• Differences occur in
• Sea level pressure
• Surface winds
• Precipitation
• Sea surface salinity ?
• Sea surface temperature
• Thermocline depth
• Sea level

Measured by satellite
6
Typical January-March Weather Anomalies and
Atmospheric Circulation
Moderate to Strong El Niño
Moderate to Strong La Niña
7
North Atlantic Oscillation (NAO)
Impact in winter
Negative NAO
Positive NAO

• More frequent stronger storms
• Warmer wetter Europe
• Colder drier Greenland
• Milder wetter US East Coast

• Fewer weaker storms
• Colder dryer Europe
• Warmer wetter Greenland
• Colder dryer US East Coast

8
In Situ SST Data

• SHIP 1800-present
• Uninsulated bucket (common before WWII)
• Insulated bucket (common after WWII)
• Engine intake (common after WWII, roughly 50)
• Hull contact (relatively new rare, most
  accurate)
• BUOY 1979-present
• Moored (thermistor and hull contact, more
  accurate)
• Drifter (hull contact)
• Biases (differences between observation truth)
• Uninsulated biased cold due to evaporation
• Engine intake biased warm due to engine room

9
Decadal In Situ SST Coverage
Number of months per decade with at least one
observation in 2o grid cell Range 0-120
Scaled 0-100 In Plot
10
Ship Buckets for SSTs
1.

• Ships wooden bucket, 1891
• Courtesy of Scottish Maritime Museum
• Other buckets
• German metal leather bucket
• UK Met Office, black insulated bucket
• UK Met Office Mk II canvas sea bucket

2.
Pictures courtesy of David Parker
11
Tropical Moored Buoy Array

• Buoy measures
• Surface wind speed and direction
• Air temperature
• SST
• Ocean temperature with depth
• Data available in real time via satellite

12
In Situ Satellite Analyses1982-present

• In Situ Data (ship buoy)
• Satellite IR Advanced Very High Resolution
  Radiometer (AVHRR)
• Began November 1981
• Satellite microwave Tropical Rainfall Measuring
  Mission (TRMM) Microwave Instrument (TMI)
• Began December 1979
• Balance between high coverage of satellite and
  ground truth of in situ data
• Climatologists Hate Biases

13
Ship Buoy Data
Week of 2-8 January 2000

• Buoy distribution complements ship distribution
• Few observations south of 40oS

14
Satellite Infrared (IR) SST Data

• Advantages
• Improved coverage over in situ data
• Multiple channels available to eliminate
  atmospheric water vapor
• High resolution 4 km
• Problems
• View must be cloud free
• Cold biases with cloud or volcanic aerosol
  contamination

15
Satellite AVHRRData
Week of 2-8 January 2000

• Missing data due to cloud cover
• Missing daytime data north of 40oN due to orbit
  of NOAA-14

16
Satellite Microwave SST Data

• Advantages
• Improved coverage over in situ data
• Improved coverage over satellite IR data
• Error characteristics different from IR data
• Problems
• View must be precipitation free
• Lower resolution than IR (50 km vs. 4 km)
• Wind speed dependent

17
Satellite Microwave TMI Data
Week of 2-8 January 2000

• Data limits of 38oS to 38oN due to orbit
• Additional diurnal limits due to time of orbit
• No impact of cloud cover

18
Comments on Errors (1)

• There are three types of errors
• Sampling
• Random
• Ship 1.3oC
• Buoy 0.5oC
• Day Satellite 0.5oC
• Night Satellite 0.3oC
• Bias differences between observation truth
• Satellites can help with
• Sampling Random errors
• Not biases

19
Schematic profiles
Comments on Errors (2)

• Profile A Nighttime or Daytime with moderate to
  strong winds
• Profile B Daytime with light (
• T(1) is the surface skin temperature
• IR satellite
• T(2) is the bulk temperature just below the skin
• Buoy
• T(3) is the bulk mixed layer temperature.
• Ship

20
SST Analyses

• Optimum interpolation (OI)
• Requires error statistics for data analyses
• Weights data by distance and relative error
• Best with lots of data
• Preliminary bias correction of satellite on
  relatively coarse spatial scales

21
SST Data and Analysis for 1 Week2-8 January
2000 La Niña
22
Objective comparisons

• Withhold a random 20 of buoy SSTs from analyses
• Define set as buoy IDs ending in 4 or 9
• Use withheld buoy data as independent data
• Compute weekly OI analyses
• WITH WITHOUT satellite bias correction
• Using different satellite data
• AVHRR only
• TMI only
• and TMI AVHRR

National Climatic Data Center
23
Weekly OI Analyses

• OI Analyses use
• AVHRR only
• TMI only
• and TMI AVHRR
• Upper panel OI without satellite bias correction
  
• Lower panel OI with satellite bias correction
• NOAA-14 biases at end of life mid 2001

Vertical axis SST anomaly Horizontal axis time
24
Bias of OI Independent Buoy Data

• OI Analyses use
• AVHRR only
• TMI only
• and TMI AVHRR
• Upper panel OI analyses without satellite bias
  correction
• Lower panel OI analysis with satellite bias
  correction

Vertical axis SST difference Horizontal axis
time
25
SST Anomaly Movie (1997-1998)
Jan 5 1997
Apr 30 1998
Dec 30 1998
Sep 5 1997
26
SST Anomaly Movie (1997-1998)
By Dave Wuertz
Jan 5, 1997 - Dec 30, 1998
Apr 30, 1998
Jan 5, 1997
Sep 5, 1997
Dec 30, 1998
27
SummarySatellite SST for Climate

• Satellite data greatly reduces SST sampling and
  random errors over in situ data alone
• Satellites must be corrected for SST bias errors
• Correction done by use of in situ data
• Because in situ data are relatively sparse,
  scales of bias corrections are limited to
  
• 5o spatial scales
• Weekly temporal scales

National Climatic Data Center