NOAA Satellite data
Like the GMS satellite , the NOAA Series of polar orbiting satellites (operated by NOAA, the US National Oceanic and Atmospheric Administration) are a component of the World Weather Watch meteorological satellite network.
Unlike GMS, however, these satellites operate in polar orbits, which means they orbit the Earth in an almost north–south direction, passing close to both poles. The orbits are circular, with an altitude between 830 km (for “morning” spacecraft) and 870 km (for “afternoon” spacecraft), and are sun synchronous. The “morning” satellites cross the equator at about 0730 local time and the “afternoon“ satellites at about 1340 local time. Accordingly, each NOAA satellite passes over the same location twice each day – once during daylight hours, and once at night. NOAA satellites orbit the Earth 14 times each day. The orbital period of the satellite is approximately 102 minutes.
Advanced Very High Resolution Radiometer (AVHRR)
One of the key instruments on board each satellite is the Advanced Very High Resolution Radiometer (AVHRR). The AVHRR is a radiation-detection imager that measures radiation reflected from and emitted by the Earth – both the surface and clouds. This radiation is measured in five (or six) different spectral, at 1.1 km ground resolution (just below the spacecraft) and up to 1350 km on either side of the orbital track. The exact details of the spectral intervals that the AVHRR measures vary from satellite to satellite, but the following table indicates general characteristics.
| Channel | Spectral interval (µm) | Resolution at nadir (km) | Typical use |
|---|---|---|---|
| 1 | 0.58–0.68 | 1.1 | Daytime cloud and surface mapping |
| 2 | 0.725–1.00 | 1.1 | Daytime detection of land / water boundaries |
| 3a | 1.58–1.64 | 1.1 | Daytime detection of snow and ice |
| 3b | 3.55–3.93 | 1.1 | Fire detection, night time detection of cloud |
| 4 | 10.30–11.30 | 1.1 | Cloud mapping, sea surface temperature |
| 5 | 11.50–12.50 | 1.1 | Sea surface temperature |
Prior to the launch of NOAA15, most earlier satellites had five-channel AVHRRs (they did not include channel 3a), but from NOAA15 onward all six channels are included, although only five channels are actually observed simultaneously. Usually channel 3a is turned on during the daylight portion of the orbit, and channel 3b during the night portion of the orbit.
NOAA satellites have the capability to both store AVHRR data onboard the satellite, for download at a Command And Data Acquisition ground station (e.g. at Wallops Island, Virginia), and to transmit the data as they move along their orbit – the latter is called High Resolution Picture Transmission (HRPT). Data stored on the satellite have degraded spatial resolution (4 km), but the HRPT data are transmitted at full 1.1 km resolution.
The sample image shows a full HRPT resolution pass over the New Zealand region. In this case the satellite was moving from south to north (passing almost directly overhead), so the southern portion of the orbit is at the top of the image. These data are for channel 4, and there are 2048 picture elements (pixels) across the full-size image and 5701 lines of data along the track of the satellite.
NIWA receives and processes all HRPT transmissions visible from its reception site at Maupuia, on top of a ridge adjacent to our Greta Point campus in Wellington, New Zealand.
After reception, the data are transmitted over a microwave link to our Greta Point campus where they are processed into data products. The coverage available from this reception site is illustrated in the adjacent image, which shows a typical set of passes over the New Zealand region.
SST Retrieval Science
The raw AVHRR data received at NIWA do not provide a direct measurement of the sea surface temperature. The data are contaminated by cloud, and atmospheric water vapour.
Accordingly, the retrieval of SST data from AVHRR data is a two stage process:
- Cloud detection – all AVHRR pixels affected by cloud (up to 90% or more in a pass) must be identified and excluded from the SST retrieval step. The approach used at NIWA is an optimal method unlike any used elsewhere in the world.
- Water vapour correction – once the clear pixels have been identified an SST may be “retrieved” from the channel 4 and 5 (and sometimes channel 3b) observations. At NIWA we utilise the SST retrieval equations derived by the NOAA/NESDIS, which are adjusted to represent 1 m depth temperatures.
There are three more steps in the generating SST products:
- After retrieval the data are “mapped” onto a standard geographic map projection, so that they can be located on the Earth’s surface.
- Because up to 90% (or more) of a single pass over the New Zealand region may be obscured by cloud it is necessary to composite the data from the individual passes over time. The fundamental idea is that clouds move, so a location that is cloudy today may not have been cloudy yesterday. Accordingly it is possible to fill in the data gaps arising from cloud by looking back through time. If this look-back time period (called the compositing period) is too short then there will still be large areas affected by cloud, but if it is too long, then the SSTs may well have changed over the compositing period. Generally, we use a three-day compositing period, and the final analysis is the average of all the clear SSTs at each location on the analysis map area.
- Lastly, the retrieved SSTs must be validated against in-situ measurements (e.g. drifting and fixed buoys) and their accuracy determined.
SST Accuracy
Comparisons between the SSTs retrieved from the satellite data and those observed by in-situ buoy observations indicate that under normal conditions, composited SSTs should have an accuracy of approximately ±0.2°C. There are two situations under which this may not be true, however:
- If the SST changes during the compositing period – perhaps as a result of strong winds from the passage of a weather front stirring the upper layer of the ocean – then the SST analysis will likely show the mean temperature of the sea surface (i.e. from before and after the stirring) rather than the current SST.
- Small satellite navigation errors (which cannot be corrected) may lead to errors in locating SST features. These should not be larger than about ±2 km.
SSTs very close to the coast may also be in error, due to the strong mixing processes that occur there (by tides, for example) and minor land contamination of the raw satellite data.
Further Information
The following papers may provide useful further information on the processes used to retrieve SSTs for the New Zealand region, the accuracy of the results, and the oceanography of the upper ocean over the south west Pacific region.
Uddstrom, M.J.; Gray, W.R.; Murphy, R.; Oien, N.A.; Murray, T. (1999). A Bayesian cloud mask for sea surface temperature retrieval. Journal of Atmospheric and Oceanic Technology 16, 117–132.
Uddstrom, M.J.; Oien, N.A. (1999). On the use of high resolution satellite data to describe the spatial and temporal variability of sea surface temperatures in the New Zealand Region. Journal of Geophysical Research (Oceans) 104(C9), 20729–20751.
