However, observing the surface of the Earth from space is not an easy job. To ‘see’ the surface, the AATSR has to look down through the atmosphere from its position high up in space. This is a bit like using our eyes to make out the features at the bottom of a muddy river – the atmosphere distorts what we are trying to measure. To complicate matters further, the temperature signal from the Earth’s surface is also distorted by the nature of the surface itself: mountains, vegetation and buildings are just some examples of what affect the temperature that the AATSR observes. To remove these effects, scientists must have some knowledge of the atmospheric and surface conditions at the time when the AATSR passes overhead. This is often a very difficult process as both the land and atmosphere vary enormously around the globe and with time. The AATSR manages to achieve the correction for the surface conditions by using ground and different satellite observations made over many years (a climatology data set). The correction for the atmosphere is estimated by using a combination of measurements made over different wavelength bands (channels), which respond differently to the properties of the atmosphere and thus provide information about its state.

To check the accuracy of the correction for the atmospheric and surface effects, the AATSR land surface temperature data are compared with land surface temperature data recorded on the ground at the same time as the AATSR makes its measurements. This process is known as ‘validation’. As these ground data are collected at the surface, where the effects of the atmosphere are not a problem and the land surface type is well known, these data should be more representative of the true surface temperature. Therefore, the difference between the ground and AATSR land surface temperature can tell scientists the accuracy of the satellite data.

Collecting ground data in Morocco

As conditions on Earth are so variable, it is very important that the validation experiments are performed in as many places as possible. If the data are found to be accurate in the Australian outback, this does not mean that we will obtain the same accuracies over the Amazon rainforest. Furthermore, it is also important that ground validation data are collected over long periods of time to determine the accuracy of the data through each season, as the atmospheric and surface conditions vary throughout the year. Although immensely useful, specific validation experiments typically yield only a few data points and a ‘snapshot’ of the accuracy of the AATSR land surface temperature data. The longer-term validation experiments provide a more complete picture and more data points, which give increased confidence in the results. The results of these validation experiments also help scientists improve the atmospheric and surface corrections, and thus the quality of the final data.

For the AATSR, a number of these LST validation experiments have taken place since the launch of Envisat. Campaigns to date include some exotic locations that include Morocco, Australia and Spain. Luckily for the scientists involved in these campaigns, the data must be collected in cloud-free conditions, as the AATSR cannot ‘see through’ clouds and so cannot measure land surface temperature when it’s raining! In general, the results of these experiments are extremely encouraging and demonstrate that, in most cloud-free cases, the AATSR land surface temperature measurements are accurate to within 1-3 K. Further improvements are also possible and efforts are ongoing to obtain the best possible performance.