GENERATION OF RAPID CIVIL ALERTS BY SATELLITE ON-BOARD SAR PROCESSING

Generation of Rapid Civil Alerts by Satellite On-Board SAR Processing

Stefan Wiehle, Anja Frost, Dominik Günzel, Björn Tings, Helko Breit, Ulrich Balss, Srikanth Mandapati

German Aerospace Center (DLR), Remote Sensing Technology Institute, Stefan.Wiehle@dlr.de, Germany

Abstract

Since decades, Synthetic Aperture Radar (SAR) satellites provide valuable information about the maritime situation. With a growing number of Earth Observation satellites the amount of acquisitions available in a certain area per time drastically increases. However, users need fast access to the retrieved information to fully benefit from this development. For maritime applications, this is especially important as acquisitions over the oceans may not only have a longer than average delay until transfer to a ground station, but also the retrieved information – ship positions, wind speeds and wave height – are very time sensitive and become deprecated within minutes.

Bringing the image generation, processing and delivery onto a satellite can save the flight time to the ground station and allow direct product delivery to users equipped with appropriate receivers. A delivery delay of just 3-4 minutes after the acquisition will enable these products to become a de-facto-standard for maritime operations, enhancing ship safety and security.

SAR processing is one component of a larger prototype system being developed in the frame of the H2020 EO-ALERT project, also comprising an optical data chain, data compression/encryption, and delivery. The system employs multiple boards with Multiple-Processor-System-On-Chip (MPSoC) combining FPGAs and ARM CPUs, which allows performing on-board processing despite the power and space constraints on a satellite. A tailored workflow and adapted L1 and L2 processing algorithms ensure that the requirements for SAR latency and product quality are met.

For L1 processing, the quad-core ARM Cortex-A53 is in charge of computing the necessary focusing, calibration, and annotation parameters based on attitude, orbit, and instrument settings on input, while the programmable logic performs all SAR signal processing such as raw data correction, pulse compression, FFTs, pixel-wise complex filter multiplications, detection, and multi-looking.

The L2 products generated within the EO-ALERT project either provide ship detection or extreme weather detection information. For ship detection, the initial Constant False Alarm Rate (CFAR) algorithm was ported to the programmable logic, while further processing steps like land masking and filtering are executed on the ARM processor. In case of extreme weather, all processing steps are performed on the ARM as the computational effort is just low enough.

While certain compromises had to be made w.r.t image quality compared to processing on a ground station, these result only in minimal effects on the final application products, which are still of high quality and provide valuable low-latency information to the end user.