Sea Surface Temperatures
Derived from NOAA Satellite Data

Introduction | Single NOAA orbits | Mosaics | SST Products | Methodology | Quality Monitoring References | Browse SST Archive

Introduction

Sea Surface Temperatures (SSTs) maps derived from remote sensing by satellites have been available since the 1970s. The Bureau currently uses measurements from the Advanced Very High Resolution Radiometer (AVHRR) on board the National Oceanic and Atmospheric Administration NOAA series of polar orbiting satellites to derive SSTs for the Australian region. The data is calibrated and quality controlled against SST data collected from ships and drifting buoys. The SSTs are used in real time operations and also archived as the data as part of Australia's National Climate Record.

NOAA Orbits

SSTs are calculated using the HRPT data received at Melbourne, Perth and Darwin for each orbit of NOAA 15, 14 and 12. The data for orbits received at Perth and Darwin will also be incorporated soon. The maximum resolution of the pixels in each orbit is 1.1 km². The SSTs for any individual orbit will have gaps where the pixels have been tested and rejected from the calculations due to suspected cloud contamination or where the satellite zenith angle is greater than 53°. Corrections are applied in the SST algorithms for intervening atmospheric absorption and to day time algorithms for reflected solar radiation. The examples of a north bound orbit below shows the reduced area of coverage of the SST algorithms for a single orbit due to cloud contamination.

Mosaics

A running 15 day composite SST mosaic in Mercator projection is used to provide complete coverage of the Australian region. The Mercator mosaic has a resolution of 2 x 2 km² at the equator increasing to 1.4 x 1.4 km² at 45°S. The latest available data pixels are used, however where pixels are rejected on the basis of cloud contamination over a sequence of orbits the data from previous days orbits are used. Areas of missing data in the composite mosaic indicate areas of cloud contamination persisting for more than 15 days. An associated age of data mosaic is also produced to complement interpretation of the SST mosaic. The samples below show the Mercator mosaics over south west WA and the Australian region.

WA region 15 day composite of SSTs from NOAA orbits

Age of SST data used in composite image

Australian region 15 day composite of SSTs from NOAA orbits

Age of SST data used in composite image

A browse service (1:5 resolution) is available to search the historical archive of satellite derived Daily Sea Surface Temperature data available from 30 December 1998.

Products

Daily regional contour maps of SSTs are produced daily by the National Meteorological Operations Centre (NMOC) and available as part of the Marine SST Service

The charts are displays of a 0.25 degree (2-5km) gridded analysis field produced by blending the satellite derived SSTs with other SST data from ships and buoys.

Grids of high resolution satellite derived SSTs in Mercator projection are also available as a subscription service on request

Global and US SST Products

NOAA NESDIS SSTs

Methodology

The AVHRR measures radiation over a number of finite channels. For any radiance measured within a specified wavelength window, there is an associated temperature such that at that temperature a blackbody would emit the same radiation. This is called the brightness temperature. The emissivity of an object is the ratio of the amount of radiation emitted by the object to that of a blackbody at the same temperature and wavelength. Therefore knowing the brightness temperature and the emissivity of the ocean surface allows the determination of the sea surface temperature.

Channels 1 and 2 are output in percent albedo using pre-launch calibration values. Using measurements of deep space (3 K) and internal blackbodies, whose temperature is measured with precision thermistors (300 K), data in channels 3, 4, and 5 are calibrated and output in terms of brightness temperature.

There are a series of checks and corrections necessary in using radiance measurements from orbiting platform to estimate sea surface temperature.

Exclusion of Data at Large Zenith Angles

Experience has shown that the sea surface temperature retrievals degrade as the sensor zenith angles increases - as we look further from nadir. The first step in the sea surface temperature retrieval is to exclude pixels that are greater than 53° from nadir.

Cloud Clearing

Measurements over cloud areas are not used (cleared ) as the cloud obscures the the sea surface and the satellite measures cloud top brightness temperatures rather than SSTs. A cloud clearing test process is used to discard any pixel contaminated by cloud from the SST calculation. A number of methods are used in different combinations for night time and day time orbits.
Test for daytime/night time
If the solar zenith angle is less than 75° and the channel 2 albedo is greater than 1% then the day time combination of tests are used, otherwise the night time combination are used.

IR uniformity test

The presence of sub-pixel cloud contamination can change the apparent brightness temperature. If we assume the SST is slowly varying, then a large variance in a local region may indicate the presence of clouds. In the first test a 3 x 3 pixel region around the pixel under consideration is used. The pixel value within the 3 x 3 region must be within 0.2°C of the median value for grid and the maximum and minimum values must be within 0.4° otherwise the centre pixel is flagged as cloud contaminated and rejected. Used in night time tests only.

Maximum Value in the Channel 2 Albedo

A large albedo in channel 2 can indicate the presence of clouds. If the albedo of a pixel exceeds 10% then the pixel is rejected. This test is used for daytime checks only.
VIS uniformity test
A large local variation in the albedo in Channel 2 also indicates clouds. All pixels within the 3 x 3 pixel region must be within 0.32% of the median value and maximum and minimum values must be within 0.64 otherwise the centre pixel is flagged as cloud contaminated and rejected. Used for day time checks only

Difference in Channel 3, 4 and 5

Channel 3,channel 4 and channel 5 have significantly different absorption responses to water vapour. A large difference in the brightness temperatures between any two of these channels can therefore indicate cloud contamination. The test against channel 3 is used at night time for test for low stratus however it is not used during day time as channel 3 also contains reflected sunlight. The test between 4 and 5 can be used for both day or night time checks.

Minimum Channel 4 Temperature

If the channel 4 temperature is too low (i.e. below 5 °C) cloud contamination is assumed and pixel rejected. Used for both day and night time checks.

SST Algorithms

If the solar zenith angle is less than or equal to 75° then the day time algorithm is used. Otherwise the night time algorithm is used.

The mean of 3 separate multi channel algorithms are used to compute the SST at night time and three algorithms must be within 2 °C otherwise the pixel is rejected. Only one algorithm is used during the day as channel 3 contains reflected sunlight and can not be used.

Daytime Algorithm

• check satellite zenith angle is less than 53°
• do land/ sea test
• check solar zenith angle - do not process if less than 1°
• gross IR test - if the channel 4 temperature is less than -5° C then do not compute a SST
• visible cloud threshold test - if the corrected albedo (albedo value divided by the cosine of the solar zenith angle) is greater than 10 percent then do not compute a SST
• visible vegetation threshold test - if the vegetation index (corrected channel 2 albedo divided by corrected channel 1 albedo) is greater than 0.75, do not compute a SST
• visible uniformity test - the corrected channel 2 albedos for all pixels in a three by three pixel box centred on the target pixel must be within 0.32 percent of the median value for the box. The maximum and minimum values must be within 0.64.
• daytime IR cloud test - if a calculated channel 4 temperature based on the channel 5 value (channel 5 temp * 1.0439 - 11.49) differs from the actual channel 4 temperature by more than 1.0° C then do not compute a SST. This test is not used by McClain et al. (1984) for daytime data but has been used historically at the BoM. The band widths for NOAA-11, 12, 14 and 15 channels 4 and 5 are the same on each satellite so the coefficients for this equation remain constant.

Multi channel SST calculation

• NOAA-11
  

CPSST Day Split Window Algorithm

sst = (0.19069 * T₅ - 49.16) / (0.20524 * T₅ - 0.17334 * T₄ - 6.78) *
(T₄ - T₅ + 0.7890) + 0.92912 * T₅ + 0.81 * (T₄ - T₅ ) * (sec(ZA) - 1) + 18.98

• NOAA-12
  

MCSST Day Split Window Algorithm

sst = (1.008574 * T₄ ) + 2.452585 * (T₄ - T₅ ) +
0.823990 * (T₄ - T₅ ) * (sec(ZA) - 1) -275.717 + 273.16

• NOAA-14
  

MCSST Day Split Window Algorithm

sst = (1.017342 * T₄ ) + 2.139588 * (T₄ - T₅ ) +
0.779706 * (T₄ - T₅ ) * (sec(ZA) - 1) + -278.43 + 273.16

• NOAA-15
  

MCSST Day Split Window Algorithm

sst = (0.959456 * T₄ ) + 2.663580 * (T₄ - T₅ ) +
0.570613 * (T₄ - T₅ ) * (sec(ZA) - 1) + -261.03 + 273.16

where:

sst - computed SST value in degrees (°) C.

T₄ - channel 4 scene temperature

T₅ - channel 5 scene temperature

ZA - solar zenith angle

Algorithms for NOAA-12, 14 and 15 posted on NOAA/ NESDIS web server. The NOAA-11 algorithm is historical and is no longer operational.

Night time Algorithm

• check satellite zenith angle is less than 53°
• do land/ sea test
• check solar zenith angle, if greater than 75° and if channel 2 reflectance is less than one percent, then use night-time algorithm. If the reflectance is greater than one percent, do not compute a SST
• gross IR test - if the channel 4 temperature is less than -5° C then do not compute a SST
• IR uniformity test - the channel 4 temperatures for all pixels in a three by three pixel box centred on the target pixel must be within 0.2° C of the median value for the box. The maximum and minimum values must be within 0.4° C
• nighttime IR cloud test - if a calculated channel 4 temperature based on the channel 5 value (channel 5 temp * 1.0439 - 11.49) differs from the actual channel 4 temperature by more than 1.0° C then do not compute a SST.
• night time low stratus cloud test - the difference obtained when subtracting the channel 3 temperature from the channel 5 temperature must be less than or equal to -0.6° C.
• Multi channel SST calculation. Three separate algorithms are used. The computed values for the three algorithms must all agree within 2.0° C.

• NOAA-11

MCSST Night Dual Channel Algorithm

sst1 = (0.17079 * T₄ - 58.47) /
(0.17334 * T₄ - 0.07747 * T₃ - 33.74) *
(T₃ - T₄ - 6.440) + 0.98530 * T₄ + 1.97 * (sec(ZA) - 1) + 15.88

MCSST Night Split Window Algorithm

sst2 = (0.19596 * T₅ - 48.61) /
(0.20254 * T₅ - 0.17334 * T₄ - 6.11) *
(T₄ - T₅ + 1.4600) + 0.95476 * T₅ + 0.98 * (T₄ - T₅ ) * (sec(ZA) - 1) + 9.32

MCSST Night Triple Channel Algorithm

sst3 = (0.16835 * T₄ - 34.32) /
(0.20524 * T₅ - 0.07747 * T₃ - 20.01) *
(T₃ - T₅ + 14.86) + 0.97120 * T₄ + 1.87 * (sec(ZA) - 1) - 3.43

• NOAA-12

MCSST Night Dual Channel Algorithm

sst1 = (1.017736 * T₃ ) + 0.426593 * (T₃ - T₄ ) +
1.800916 * (sec(ZA) - 1) - 276.264 + 273.16

MCSST Night Split Window Algorithm

sst2 = (1.013674 * T₄ ) + 2.443474 * (T₄ - T₅ ) +
0.314312 * (T₄ - T₅ ) * (sec(ZA) - 1) - 277.797 + 273.16

MCSST Night Triple Channel Algorithm

sst3 = (1.003194 * T₄ ) + 1.007171 * (T₃ - T₅ ) +
1.174698 * (sec(ZA) - 1) - 273.262 + 273.16

• NOAA-14
  

MCSST Night Dual Channel Algorithm

sst1 = (1.008751 * T₄ ) + 1.409936 * (T₃ - T₄ ) +
1.975581 * (sec(ZA) - 1) - 273.914 + 273.16

MCSST Night Split Window Algorithm

sst2 = (1.029088 * T₄ ) + 2.275385 * (T₄ - T₅ ) +
0.752567 * (T₄ - T₅ ) * (sec(ZA) - 1) - 282.24 + 273.16

MCSST Night Triple Channel Algorithm

sst3 = (1.010037 * T₄ ) + 0.920822 * (T₃ - T₅ ) +
0.067026 * (sec(ZA) - 1) - 275.364 + 273.16

• NOAA-15

MCSST Night Dual Channel Algorithm

sst1 = (1.041037 * T₄ ) + 1.587582 * (T₃ - T₄ ) +
1.677430 * (sec(ZA) - 1) - 283.51 + 273.16

MCSST Night Split Window Algorithm

sst2 = (0.993892 * T₄ ) + 2.752347 * (T₄ - T₅ ) +
0.662999 * (T₄ - T₅ ) * (sec(ZA) - 1) - 271.40 + 273.16

MCSST Night Triple Channel Algorithm

sst3 = (1.015354 * T₄ ) + 1.063572 * (T₃ - T₅ ) +
1.294955 * (sec(ZA) - 1) - 276.76 + 273.16

where:

sst n - computed SST value in degrees ° C.

T₃ - channel 3 scene temperature

T₄ - channel 4 scene temperature

T₅ - channel 5 scene temperature

ZA - solar zenith angle

Algorithms for NOAA-12, 14 and 15 posted on NOAA/ NESDIS web server. The NOAA-11 algorithm is historical and is non-operational.

• the SST value used is the mean of the three values computed.

Climatology test

Computed SST rejected if differs from climatology by more than 10°

Quality Monitoring

Verification of the computed SST fields is carried out automatically after each individual orbit is processed. SST temperatures are compared to ship and buoy sea-temperature observations which are co-located in both space and time. Observations within 3 hours of the nominal AVHRR image time are accepted.

The results of the comparisons may be displayed both graphically and in tabular form and are accessible via the Bureau internal web. Statistics for the test system have been kept since July 1999.

References

McClain, E. P, et al., 1984, 'Comparative Performance of AVHRR-Based Multichannel Sea Surface Temperatures', J. Geophys. Res., 90, pp11587-11601

NOAA Polar Orbiter Data User's Guide

• http://www2.ncdc.noaa.gov/docs/podug/index.htm

• http://www2.ncdc.noaa.gov/docs/klm/

• http://manati.orbit.nesdis.noaa.gov/sst/cwIntroduction.html

Links valid at: 26th July 2007