Space | 19. October 2010

Baptism of fire for the satellite formation

Last Thursday, the two German radar satellites of the TanDEM-X formation finally reached their operational orbit configuration, with only a few hundred metres separating them. In this configuration, they act as a unique radar interferometer in space. The next day, on Friday evening, the instruments were switched on, after many careful checks, to acquire the world's first Digital Elevation Model (DEM) data using a free-flying bistatic Synthetic Aperture Radar (SAR) satellite formation. That same night, the jointly acquired data were received and processed by our operational processing chain.

Bistatic acquisition means that both instruments work simultaneously in a synchronised way. Their data streams have to be aligned in a complex way during on-ground processing, in order to derive interferometric data for DEM generation. This was an absolute first for us. Although we have already generated thousands of DEMs over the past few months, they were all acquired using the 'classical' technique of two data acquisitions of the same area but separated in time. We used bistatic test data acquisitions from our colleagues (they blogged about them) to test bistatic processing, but the phase information needed for DEM generation was not available in these data, acquired when the satellites were 20 kilometres apart.

The very first bistatic DEM acquisition captured a wonderful scene on Sicily's east coast, with the Mount Etna volcano right in the centre. Processing of both data sets worked flawlessly. The two matching bistatic images were studied and celebrated by the team. By noon on Saturday, we had received many additional bistatic data sets at Neustrelitz and processed them to derive DEMs here in Oberpfaffenhofen. This was an outstanding achievement for the entire ground segment.

The PGS SAR Team with the first bistatic DEM acquisition image pair.

TanDEM-X-Höhenmodell im Vergleich zum SRTM-Höhenmodell

The bistatic TanDEM-X data is impressive in its quality even though we have just started the bistatic-commissioning phase of the mission. The final DEM data products will be generated from a combination of several acquisitions covering the same spot on the Earth to achieve the required precision of just a few metres of elevation error all over the world. Nevertheless, already these first bistatic DEMs provide very fine details of the landscape. Specifically, in comparison to the best available global data sets – which are not even of a constant quality level everywhere (for example, Shuttle Radar Topography Mission (SRTM) DEMs) – the improvements are tremendous. The step from the 30–90 metre raster of SRTM to the 12 metres of TanDEM-X may not sound too impressive, but it is in fact like switching from a VGA webcam (or, in the best cases, a two-megapixel digital camera) to one with a resolution of 12.5-megapixels.

The image on the left shows a comparison between the bistatic TanDEM-X DEM of the eastern face of Mount Etna with data from SRTM.



What is so special about bistatic interferometry?



Firstly, both instruments acquire data from an area of Earth's surface at exactly the same time – nothing can change on ground between acquisitions. Secondly, one of the two satellites can just 'listen' to the signals that come back from the ground having been transmitted by its companion; meaning that one of the two satellites can save battery power and keep its radar instrument cool. In this way, the time for data acquisition is doubled, compared to the classical monostatic approach. It sounds like nothing but advantages – however, the technique is extremely complicated.

One of the most important aspects is that both radar instruments are always working with pulses. For the timing of these pulses, they need a kind of clock, the internal oscillator. Unfortunately, the two oscillators are completely independent from each other hence they might be slightly out of step. Thus both instruments may work simultaneously, but each one uses its own 'beat'. What would be a disaster for an orchestra is actually not really a problem for the satellite duo – as long as we know the timing of each of them during processing on the ground. Therefore, synchronisation pulses are exchanged, with which each satellite records the 'beat' of its partner.

The precision required during processing to generate accurate DEMs is enormous; minimal timing differences may shift the images by hundreds of metres and – even worse – destroy the frequency and phase information that is needed for DEM generation. Without synchronisation corrections, the smallest errors would build up to wrongly calculated kilometre-high ramps in the height data. Telling you this, you might better understand how excited we were by the first faultless DEMs. They confirmed the successful implementation of this technique in all the systems in space and on the ground by the many engineers working on the project.

Bistatic DEMs – what is the advantage?

Taking these complications into account, what is the benefit of bistatic data besides the two advantages already mentioned? One important aspect has to be considered; the monostatic technique that we have already used to generate thousands of DEMs, works best in places where no vegetation or water disturbs the imaging process with their tiny movements. The new bistatic DEMs of ice and sand deserts are impressive – like the images of October Revolution Island in the Arctic Ocean or the one of Death Valley, in California … however, these areas are uninhabited.

DEM der Oktober-Revolutionsinsel

Bistatic DEM of the ice and rock formations on October Revolution Island (radar brightness image coloured using DEM height data).

DEM des Death Valley

Bistatic DEM of Death Valley, California.

In order to observe not only these very important (but isolated) regions – for example, to conduct geodynamic and climate research – but also the critical volcanic and earthquake-prone regions inhabited by humans, a more robust technique is needed – and that is bistatic interferometry. Water, forests and other vegetated areas will not be completely coherent (and are thus not free of phase noise), but they can be processed to create DEMs much better than monostatic data.

It was not by chance that several volcanoes were selected as the first bistatic targets; the investigation and monitoring of geologically active zones is one of the many tasks of the TanDEM-X and TerraSAR-X mission. Now, for example, we are able to generate accurate DEMs of the active volcano Merapi on Java, which is covered by dense vegetation and has thousands of people living on its slopes.

Bistatic DEM of the Merapi volcano on Java, Indonesia. The many bright patches on its slopes are villages and towns. Image credits: DLR.

We experienced an eventful weekend and we can conclude that the mission systems underwent their baptism of fire and are fit for the tasks of the coming months and years of the mission.