Introduction
The routing of synthetic aperture radar (SAR) imagery from the European Space Agency's ERS-1 and ERS-2 platforms is a typical example of how information is captured, distributed and processed (Figure 1). Raw digital data is downlinked to a number of ground stations strategically located around the world where it may be processed on-site or more usually transferred elsewhere for processing and archiving.
      In the case of the ERS series there are four 'primary' ESA ground stations in Europe and Canada together with more than 20 collaborating 'national' and 'foreign' stations. ESA's own resources at Kiruna (Sweden), Fucino (Italy) and Maspalomas (Canary Island) are equipped to provide customers with a 'fast delivery' service for global wind and wave products which are typically only of value if in customers' hands within a few hours. Other national facilities at Troms¿ (Norway), Gatineau (Canada) and Fairbanks (Alaska) offer similar services. But the majority of off-line processing is dealt with at Processing & Archiving Facilities (PAFs) in Germany, Great Britain and Italy. Data may be moved between centers over terrestrial or satellite links but more often by shipping the bulky high density tapes physically by road or air. Transportation costs and the ability to interchange data reliably among the various sites are therefore crucial operational issues.
      Economic considerations usually mean that current ground stations must support a number of separate spacecraft - often referred to as multi-mission support. As Table 1 demonstrates, there is little or no standardization of data rates. Many platforms use an on-board recorder to store imagery while the spacecraft is out of sight of its ground stations. This is downlinked later in a time-inverted format so that playing the ground-station tapes backwards is the most practical means of restoring the original direction of the time history.

The Need for New Recording Techniques
Conventional high density digital recorders (HDDRs) using open reel longitudinal technology continue to serve the remote sensing community well in many operational roles. Some users look forward to the day when it will be possible to process all data in real time as it is received, but in the meantime as the volume of data to be collected, processed and archived continues to grow almost exponentially, the smallest improvements in cost effectiveness and throughput can yield dramatic returns. The data recording problems facing today's station designers and operators can be summarized as:
• Improving flexibility - the need for existing sites to support an increasing number of platforms.
• Increasing throughput - the 'production line' environment and the need to improve the service to customers at the lowest possible cost.
• Technology issues - weighing the benefits of adherence to existing recording formats against the advantages of 'next generation' solutions.
• Mobile applications - balancing the desirability of compactness with the requirement for compatibility with other recorders in the network.
• Shipping costs - cycling high density digital tapes between acquisition and processing sites is a significant element of overall operating costs.
• Archiving - whether to store unprocessed data in its original form or to reduce its volume by transcription. Dealing with old archives.
• Backing up disk arrays - this technology has been proposed for 'fast delivery' processing at receiving sites but its single-pass storage capacity provides no permanent means of backing up raw (unprocessed) data and complicates the process of transferring data to the PAFs.

Micro-Track Recording
Penny & Giles Data Systems has more than 15 years' experience with global-scale recording networks and has used this background to develop a new cassette recorder known as PEGASUS intended particularly for remote sensing applications. PEGASUS employs a technology known as D-1 micro-track recording derived from the company's work with the narrow track techniques used originally by radio astronomers to increase the capacity of conventional 1-inch open reels.
      Forty parallel tracks are written along the length of a 19 mm wide tape mounted in a standard D-1 (L) broadcast video cassette. Data is buffered and written to tape at a fixed rate regardless of the user's input clock. For increased capacity, two end-to-end passes are recorded. The track width, although considerably less than that of a conventional HDDR, is sufficient to guarantee reliable data interchange. Although costing typically 35 percent less than a 15-inch open reel, a single D-1(L) cassette can hold data from three complete 10-minute 105 Megabit/sec satellite passes compared to the single pass capacity of the open reel.

Handling Multi-Mission Data
The input/output buffer solves the problem of handling the data rates from different platforms. The recorder will accept any rate up to 110 Mbits/sec without adjustment (or even higher in 'burst' mode). By routing the outputs from several platform specific bit synchronizers to one or more recorders via a computer controlled data path switch, multi-mission support can be totally automated.
      Since micro-track recording is a linear (rather than a helical) format, the reverse data downlink re-reversal process is dealt with simply by replaying the tapes in the opposite direction. However, perhaps the most important benefit is that data is written and read at a fixed tape speed and a constant packing density, eliminating the need for complex multi-speed equalization circuits and lengthy calibration procedures.

Advanced System Control and Data Management
While the micro-track recorder used in its continuous (streaming) mode is able to fit into an existing facility with minimal disruption, this is to ignore other advanced data and control features designed to improve the flexibility and performance of the overall data handling process. For example, handshaking lines in the input/output interface can be used to control the data flow to and from the recorder.
      On the record side, the source can use these lines to ensure that the recorder is ready to receive data and that only valid data is recorded on tape. One practical advantage of this is that it is not necessary to waste tape by running the recorder up to speed in anticipation of signal acquisition. During replay, the flow of data can be controlled by the processor on an 'on demand' basis with transfers taking place at a rate optimized by the processor itself. IEEE488 and RS-449 control interfaces are available as standard while the PEGASUS design anticipates operation in an 'open system' environment with the inclusion of an industry-standard SCSI-2 (fast/wide) data/control interface.
      Managing the large number of tapes in circulation within a remote sensing data capture and processing network has historically been a major concern, necessitating rigorous and time consuming manual identification procedures. An important distinguishing feature of 'new generation' recorders is the extent to which the data management can be automated. For example, PEGASUS logs time, date, the serial numbers of the recorder and cassette together with various other system information. In addition, 'event markers' can be assigned to the data, for example to tag individual satellite passes, and together this information provides a convenient and reliable means of managing cassettes and their contents. It is also possible to record and replay station timecodes such as IRIG A & B and NASA 36 synchronously with the data.
      But perhaps the most useful control aspect is the degree of 'intelligence' now placed within the recorder itself. The classical tape recorder functions RUN, STOP, FAST REVERSE, etc., are replaced by composite commands such as WRITE, READ, STOP DATA, READ FORWARD {start_event_marker, stop_event_marker} and FIND TIMECODE {date + time}. A comprehensive menu of status requests and responses (solicited and unsolicited) allows the user to check and control many system parameters concerned with the configuration and performance of the recorder and media.

Upgrading Existing Recorder Installations
At present, many equipment procurements are simply to upgrade existing installations, perhaps to replace early generation open reel systems or more often to work alongside current HDDRs. In either case, it is desirable that the surrounding hardware and software environment should remain as undisturbed as possible (although some changes may be unavoidable when new generation products are introduced). Yet users also need to be confident that new equipment will remain fully compatible with evolving data capture and processing strategies. For example, the unique flexibility of the PEGASUS input/output circuitry ensures that any standard data I/O format can be selected by simple remote commands. Bit-serial ECL and 8, 16 and 32-bit parallel inputs can all be accommodated without manual intervention, while the design allows for a SCSI-2 (fast/wide) control/data interface to be implemented when necessary. The unit's simple remote control architecture provides scope for software emulation of the user's existing command sets.

Transportable Ground Stations
The 'transportable ground station' is perhaps the most exciting development in remote sensing in recent years. The proportion of the globe which can be imaged routinely by a particular satellite operator depends to a large extent on access to ground station resources in the regions concerned. This is dictated primarily by the 'line-of-sight' requirements of low Earth-orbit polar satellites. In regions unable to provide adequate acquisition resources for effective operation, self-contained transportable facilities offer an attractive alternative.
      PEGASUS recorders have been supplied to the TELEOS program, a new generation of transportable ground stations manufactured by Datron/Transc Inc. (DTi) to acquire data in a multi-mission role anywhere in the world. Figure 3 shows a 3.6 meter TELEOS system, incorporating a single recorder, which has recently been deployed in Nairobi, Kenya to collect SPOT imagery of Eastern and Southern Africa. (TELEOS is a collaborative venture between Telespazio of Italy and EOSAT of the USA.) Figure 2 shows the interior of the operations shelter of a 13 meter transportable station manufactured by DTi for the National University of Singapore. Two compact PEGASUS recorders can be seen located near the base of the left hand equipment racks.

Transcription and Archiving
The cost of transporting magnetic tapes between receiving sites and their associated PAFs represents a substantial portion of a network's operating budget. One operator has calculated that shipping costs could be reduced by as much as 50 percent simply by transcribing data from 15-inch open reels (one 10 minute pass per reel) to PEGASUS D-1(L) cassettes (three passes per cassette) for shipment between its field receiving station and the PAF.
      In some cases the concept of archiving on cassettes may be an extension of the transcription process, particularly where operators have already elected to transport data from receiving sites in that format. But the warehousing problem itself will generally be reason enough to consider the introduction of a new technology. The amount of data produced by today's platforms is typically measured in TeraBytes (1 TeraByte = 1 million MegaBytes) posing significant storage and handling problems even for 'current' (active) data, not to mention long-term storage of the products of previous missions. In the archiving area a number of additional issues may be involved:
• Obsolescence of current equipment.
• Media aging.
• Physical space requirements.
• Reliable access to archive material.
      It might be assumed that media life would be the most critical factor governing tape-based archives but it is more often said that properly stored tapes may actually be serviceable long after the machines used to record them have been superseded by later technologies. However true this may be, media manufacturers recommend that archived material be transcribed every 10 years or so. In practice, this generally involves migrating the data from an 'old' to a 'new' recording technology. At the same time, most users take this opportunity to compress the archive physically as far as possible to save on storage costs. Lastly, it is essential that archived material can be located rapidly when required.
      The question of how data should be archived is now being tackled on several levels simultaneously. A D1(L) micro-track cassette can store up to six times more information than a 15-inch open reel yet costs maybe 35 percent less. Existing users of these systems are finding it convenient therefore not only to archive recent material in this form but also to transcribe data from earlier missions onto micro-track cassettes. An identical replay system can then be used to retrieve many classes of archive material - an inexpensive workstation being used to control the whole process. For very large archives, a robotic media handling capability can improve efficiency still further.

Back-Up for Disk Arrays
Disk array technology has been considered as a possible solution to the problem of providing both high speed data capture and fast image processing at receiving sites. Arrays can be engineered to support the high real-time acquisition rates involved. Typically, an array is used to store imagery from a single pass and process it fully before the spacecraft's next transit. This is essentially an 'on-line' operation with all the attendant risks of equipment failure and potential loss of data. Recording the raw data in parallel on a cassette provides not only a high speed real-time back-up facility for the disk array but, perhaps more importantly, offers a convenient and cost effective way of transporting the material to PAFs and other users.

Conclusion
Cassette recorders are replacing open reel HDDRs as the standard means of capturing and storing unprocessed remote sensing data. This is true both in conventional ground station environments and in the emerging transportable sector where the recorder's small size and superior flexibility are invaluable in the self-contained 'go anywhere, do anything' role.
      At first, many users will avoid changing system architecture more than necessary, preferring that the new equipment should emulate HDDRs. In time, the more sophisticated control and data interfacing features will be used to greater effect, driven by the need for improved throughput and operational cost effectiveness.
      The expense of shipping media between sites can be reduced considerably if imagery is first transcribed at higher density onto lower cost cassettes. Processing and archiving facilities can take the opportunity afforded by the new technology to migrate existing data sets into the same format to streamline the image processing element of the chain. Cassette recorders can also provide a flexible and convenient means of backing up and transporting data acquired using disk arrays.

About the Author:
Edwin Kayes is data recording marketing manager with Penny & Giles Data Systems, Wells, England.

BACK