By: Owen Embury
Oceans cover over 70% of the Earth’s surface and knowing its temperature is crucial for understanding both weather and climate. Historically, sea surface temperatures (SSTs) have been measured in situ – from ships and automated buoys – however, in recent decades we have also been able to use satellites to make observations of the Earth system. While satellites make indirect observations of the SST by measuring the thermal radiation emitted by the sea, they can provide significantly more coverage than in situ observations as a single satellite can see the whole Earth over a day or so. We now have a global, gap-free, 38-year time-series of SST generated as part of the European Space Agency (ESA) Climate Change Initiative (CCI) for SST and continued under the Copernicus Climate Change Service (C3S).
Our SST data are generated from infra-red sensors flown on-board polar orbiting satellites. These satellites circle the Earth in a low orbit (around 800 km) travelling in a north/south direction passing over each pole and taking about 100 minutes for each orbit. Each satellite will be able to see the whole of the Earth’s surface over a day or so as the Earth rotates. We use two main types of infra-red sensors to measure the SST. The first are the Advanced Very High Resolution Radiometers (AVHRRs) – a series of meteorological instruments which have been in use since the late 1970s. By modern standards, these instruments are no longer advanced or high resolution! The second set of instruments are the Along Track Scanning Radiometers (ATSRs) and Sea and Land Surface Temperature Radiometers (SLSTRs). The instruments were specifically designed to make high accuracy climate observations of the surface temperature and can be used to improve the accuracy of the AVHRR observations.
Figure 1: Global mean diurnal cycle of daily SST anomalies from drifting buoys for different wind speeds. The annual mean diurnal cycle is displayed in black, Northern Hemisphere winter (DJF) in blue, spring (MAM) in green, summer (JJA) in red and autumn (SON) in orange. The bars in the top left corner show the associated uncertainties. Reproduced from Morak-Bozzo et al. 2016.
When generating a climate data record, we are interested in how the SST changes over long time periods (years and decades); however, the SST changes throughout the day as the ocean warms during the day and then cools again overnight (the diurnal cycle). In order to separate these two effects, we provide the satellite-based temperature at a standardised time of day of 10:30 am or 10:30 pm local time (which is a good estimate of the daily average temperature). Figure 1 shows how the SST changes through the day at different wind speeds. The cycle is largest at low wind speeds where the surface warms rapidly from dawn to mid-afternoon. The cycle becomes smaller as the wind speed increases, because the wind causes the surface water to mix with lower water resulting in a more constant temperature.
Figure 2: – SST product levels. L2P data are the original satellite viewing geometry with gaps due to cloud cover and land. These are gridded to L3U, and then combined to produce daily L3C. Data from multiple sensors are combined and interpolated to produce a gap-free L4 SST.
SST data are provided on multiple “levels”, each adding a bit more processing to make the data more convenient for users. The lowest level data containing SST is known as L2P which contains the full resolution imagery as observed by the satellite. When displayed as an image, it forms a long rectangle along the satellite’s direction of travel. To make the data easier for users, we first grid it onto a global 0.05° latitude-longitude grid (L3U), where we can see the satellites orbit around the Earth. Next, all the data from one day is collated to produce L3C at which point we have near-global coverage apart from gaps due to clouds. Finally, we produce a gap-filled estimate of the daily-mean SST which combines data from multiple satellites and interpolates into the gaps to give a Level 4 analysis product. These SST product levels are shown in Figure 2.