By the time Mariner 6 and Mariner 7 were sent to Mars four years later, in 1969, Davies was a key part of the team dealing with the images they sent back. His particular contribution was to work on the mathematical techniques needed to turn the disparate images into the most reliable possible representation of the planet.
Since the seventeenth century, when Willebrord Snell of Leiden first refined the procedure into something like its modern form, earthbound map makers have turned what can be seen into what can be precisely represented through surveying. Decide on a set of landmarks – Snell and his countrymen liked churches – and then, from each of these landmarks, take the bearings of the other landmarks nearby. From this survey data you can build up a network of fixed points all across the landscape. Plot every point on your map according to measurements made with respect to things in this well-defined network and it will be highly accurate. If, unlike Holland, your country is large, mountainous and only sparsely supplied with steeples, setting up a reliable network in the first place can be hard work – the United States wasn’t properly covered by a single mapping network until the 1930s, when abundant Works Progress Administration labour was available to help with the surveying. But the principle of measuring the angles between lines joining landmarks has been used in basically the same way all over the earth.
Two problems make the mapping of other planets different, one conceptual, one practical. On earth, experience allows you to know what the features you are mapping are: hills, valleys, forests and so on are easily recognised for what they are. While the pictures a spacecraft’s cameras send back may be very good, this level of understanding is just not immediately available. When the first images of Mars were sent back by Mariner 4 they were initially unintelligible to Murray and the rest of the imaging team. Before the researchers even started on a physical map, they needed a conceptual one, a way of categorising what was before their eyes. How to do this – how to see what had never been seen before – was the besetting problem of early planetary exploration.
The practical problem is that unlike an earthly surveyor, you can’t wander around the surface of an alien planet making measurements at leisure. Your only viewpoint is that of a spacecraft flying past the surface at considerable distance and speed. So you not only don’t know what you’re looking at; you’re also none too sure of where you’re looking from. A spacecraft’s position is not a given, like that of a church. It is something that its controllers have to continuously work out. What they know for sure is how fast it is receding from the earth, because that causes frequency changes in its radio signals. To find out where the spacecraft actually is, this information is compared with estimates of where the spacecraft thinks it is – the primary tools here are small on-board cameras called star trackers – and calculations of where it ought to be, derived from measurements and models of all the forces – the gravity of the sun and the planets, the gentle nudges from on-board thrusters – that are shaping its trajectory. If all is going well, the calculations based on all these observations fall into line to produce a consistent picture.* But though this may be accurate enough for navigation, it is not accurate enough for map making. You don’t know precisely where the spacecraft is, or precisely which way its camera is pointing, or, for that matter, precisely where the surface of the planet is. So you can’t say exactly what bit of the planet you’re looking at in any given picture.
Working round these problems involved Davies in a huge amount of laborious cross-checking and number crunching (Airy himself would have loved it, I suspect). First he had to put together a set of clearly distinguishable features that appeared in more than one of the pictures – the centres of craters, for the most part. The precise locations of these features within the individual frames in the data sent back by the spacecraft then had to be put into a set of mathematical equations along with the best available figures for the spacecraft’s position when each picture was taken, and the direction in which the camera was pointing at the time. Then he had to add in factors describing the distortions the cameras were known to inflict on the pictures they took. Once all this was done, the whole calculation had to be fed into a computer on punch cards; the computer then ground through possible solutions until it came to one that made the values of all the variables in the equations consistent. Those values defined a specific way of arranging the set of surface features in three dimensions – imagine it as a framework of dots linked by straight lines – which came as close as possible to satisfying all the data. Effectively, the final answer said ‘if the reference points you’ve specified are arranged in just this way with respect to one another, and if the spacecraft was at these particular points at these particular times, then that would explain why the reference points appear in the positions that they do in these pictures’. That optimal arrangement of reference points was the control net.
Once Davies and his colleagues provided the control net, it could be used to position all the rest of the data. It became possible to say quite accurately where things were with respect to the planet’s poles and its prime meridian. Indeed, one of the primary functions of the control net was to define the planet’s latitude and longitude system – which is why Davies, as both maker of the control net and a member of the International Astronomical Union committee responsible for giving names to features on other planets, was able to put Mars’s Greenwich in a little round crater within the larger crater that was being named after Airy.
Since his first work on Mars, Davies has done his bit in the mapping of more or less every solid body any American spacecraft has visited. By the 1970s he had completely forsaken the black world of spy satellites for the scientific delights of other planets and the personal pleasure of exploring this one: once unencumbered by security clearances and the knowledge they bring, he was free to travel to meetings all around the world, and did so with Louise and alacrity. He’s never made headlines – I doubt he’d want to – but his contributions have been vital prerequisites for much of the work that has.
But there’s still more that Mert would like to do. The mathematics of the control net maximise its self-consistency, not its accuracy. This makes it likely that it contains errors. If you had some independent way of checking it – if you had a point in the control net the location of which you knew independently – you might be able to do something about that.
In principle, such independent measurements are possible. When I interviewed Davies in his office at RAND in December 1999, America had landed three spacecraft on the surface of Mars – the two Viking landers in 1976, and Pathfinder in 1997. The radio signals sent back from those spacecraft revealed their positions very accurately with respect to the fixed-star reference system used by astronomers. If you could find the spacecraft in images of the Martian surface that also contained features tied into the control net, you could check the position of the spacecraft with respect to the net against its absolute position as revealed by the radio signals. That would allow you to calibrate the net with new precision. Do the same for a few spacecraft and you could tie the thing down to within a few hundred metres, as opposed to a few kilometres.
The frustration is that you can’t see the spacecraft. About the size of small cars, from orbital distances – hundreds of kilometres – they are lost in the Martian deserts. The Mars Observer Camera, part of the Mars Global Surveyor spacecraft, has been trying to pick out some sign of the three spacecraft since 1997. It is by far the most acute camera ever sent to Mars. But even MOC can’t pick out the landers. Mankind has made its mark on Mars – but that mark has yet to be seen.
Lacking any proper sightings, checks on the control net using the landers’ locations have had to be indirect. From matching the features that the landers see on the horizon around them with features visible in pictures taken from orbit, it’s possible to make estimates of where the landers are, estimates that are potentially very accurate. Unfortunately, the different experts who try this sort of triangulation get different answers. When