Due South with the British Antarctic Survey

A cold and windy September afternoon marks the start of the fifth Atlantic Meridional Transect (AMT) experiment aboard the British Antarctic Survey (BAS) research vessel RRS James Clark Ross. Each year the ship sails from the UK to the Falkland Islands in September en route to service the UK Antarctic research bases. Jim (our UNIX support engineer) is busy fastening down his trusty Toshiba laptop (Tecra 730XCDT with 48MB) in his cabin on board James Clark Ross in preparation for the inevitable bad weather. Ahead of us lie six weeks of precision ocean-atmosphere measurements, near real-time data processing, heated debate, troubleshooting and, hopefully, some scientific discovery. Fortunately, we have both chosen one of the most versatile and reliable operating systems at hand to maximize our output during this experiment—Linux. This article discusses the impact of Linux, which is now routinely used at BAS and during the AMT experiments. (See http://www1.npm.ac.uk/amt/ for more information on the AMT project.)

Figure 1. The British Antarctic Survey research vessel RRS James Clark Ross used for the AMT experiments (Photo: Tony Bale)

The RRS James Clark Ross (shown in Figure 1) was launched in 1990 and is one of the world's most complex marine research vessels, incorporating over 400 square meters of scientific laboratory space. It was named after the scientist and polar explorer Admiral Sir James Clark Ross, RN (1800-1862), who in February 1842 reached 78.9 degrees S—a record which remarkably stood well into the 20th century.

Figure 2 shows the track taken by the James Clark Ross during the and AMT-5 cruise superimposed on a monthly composite satellite image of Chlorophyll a derived from the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) carried on the SeaStar spacecraft. Different types of phytoplankton have characteristically different concentrations of chlorophyll and are, therefore, different in color. By measuring the color of the ocean with the SeaWiFS instrument, estimates of the amount and general type of phytoplankton in specific regions can be made as shown in Figure 2. An extensive web site provides a wealth of information on the SeaWiFS mission at http://seawifs.gsfc.nasa.gov/SEAWIFS.html.

Figure 2. Ground track of the RRS James Clark Ross during the AMT-5 experiment superimposed onto a SeaWiFS satellite composite image of Chlorophyll a. White marks regions of cloud and the arrow highlights the direction of the north equatorial current system which is pushing nutrient rich Amazon water to the north west. (Image courtesy of Stanford B. Hooker.)

The ocean-atmosphere measurements taken during the AMT cruises are fundamental for the calibration, validation and interpretation of remotely sensed observations, including sea-surface temperature, wind speed, atmospheric water vapour and ocean colour (which can be related to bio-optical processes)—all of which are vital for ongoing climate research. Figure 3 shows the instrument cluster mounted on the forward mast of the James Clark Ross which measures, among other things, sea-surface temperature (using an infrared radiometer), solar radiation, wind speed and direction, air temperature, radar backscatter (a measure of surface roughness) and humidity. All of these measurements are used to investigate the processes, occurring at the air-sea interface, which in many cases define the signal actually measured by the satellite instruments. (See http://www1.npm.ac.uk/amt/ for more information on the AMT project.)

Figure 3. Forward mast installation of the RRS James Clark Ross showing some of the radiometers and the radar system.

From a system engineer's point of view, the main requirement of an operating system is that it integrate with the existing computing infrastructure available on (in this case) BAS ships and also at Antarctic bases as well as at headquarters. Since the majority of instruments are logged to Sun SPARC workstations, it makes sense to run an OS which allows NFS mounts to the data areas the workstations create. Many users require the ability to perform data processing locally without having to place extra strain on the data-logging workstations. As most of this is undertaken using shell scripts or compiled C source code, it made sense to run a form of UNIX locally. This enables shell scripts to run with no modifications and the C code to easily recompile. Other user-driven considerations included access to backup hardware, hard-copy output, real-time data displays and access to a common library of software packages.

Figure 4. The main laboratory space occupied by the Linux workstations aboard the RRS James Clark Ross during the AMT experiments.

Upon considering these requirements, our choice was made simple: it had to be Red Hat Linux. For our purpose, Linux provides an extremely versatile operating system with the ability to effortlessly and seamlessly integrate itself efficiently into any existing network system. The immense amount of supported hardware made installation on Jim's laptop and on our desktop machines a painless exercise, and we were delighted with the way that even a default installation gave us exactly what we wanted. It is this kind of hardware support and user-friendly installation interface which caught our eye in the first place when considering various UNIX systems for Intel processors.

After unpacking our brand new PC (Intel DX 120; 32MB RAM) and removing the pre-installed OS (we did ask for Linux), the whole installation took less than an afternoon. We now had a fully working UNIX workstation configured within the James Clark Ross NIS domain, auto-mounting file systems whenever required. Jim went even further with his laptop. Using Caldera Wabi 2.0, he had the ability to run the BAS standard word processor and e-mail packages (which are MS Windows-based). Being able to do all this using free or inexpensive software proves what a professional product Linux is. Gone are the days when Linux was considered a “toy UNIX” for hackers—it is now a fully functional UNIX environment which is just as stable (if not more) as the various commercial UNIX systems on the market.