How I created a Martian map and why

Mars, including the very extent of our map, has been mapped a number of times before; sometimes in splendid topographic detail such as those made by Technische Universität Berlin for the European Space Agency (ESA) nine years ago.

So a common question has been ‘why have we bothered to map it also?’ I have been interviewed by a number of media outlets this week asking about the map. The press coverage has been fantastic – leading to 37,000 views of the map sheet on Flickr in the first 24 hours – but using my own blog I hope to share a more in-depth summary with you.

I was approached by Peter (Dr. Grindrod) who happens to like OS maps and read this blog. He had this idea of recreating Martian mapping as a more formalised map sheet in keeping with the house style of Ordnance Survey (OS). An exciting opportunity, space has always interested me; oddly enough Planetary Cartography is something I failed to mention in last year’s AGI Foresight 2020 report (apologies to those in that field of mapping).

The original intention was to create a map with an overlay of traverses and sites of interest to showcase the ability of a map designed by a dedicated cartographer to illustrate events on and stories from Mars.

A purely fictional traverse and landing site above the map
A purely fictional traverse and landing site above the map

A lot of planetary data is open data but the detailed thematic information might not always be so, therefore for external publication we decided to publish, via social media and in conjunction with The Times newspaper, the basemap alone, free from any overlay.

Although the power of a map is its ability to tell a story, the thematic overlay is seldom a problem at OS because we have a tendency to style the base of our map subtly enough to leave room in the primary end of the colour spectrum for more obvious overlay themes.

So my main mission became to create a base map befitting of an OS paper map sheet and retaining as much of the glorious detail in the National Aeronautics and Space Administration (NASA)’s elevation data as possible.

The map sheet, ‘Western Arabia Terra’, is a basemap in typical OS-style showing an area of Mars similar in size to the USA, China, Canada or Brazil. In terms of cartography, I completely designed the match from scratch but with our traditional paper mapping in mind. Although the scale was more in line with a typical OS Wall Map product, having a section of such detailed elevation data led me to believe that the style of a more detailed map sheet such as an OS Explorer Map or OS Landranger Map might be more fitting and would add to the OS look and feel of the map.

Peter was an enormous help in supplying me with ArcGIS-ready data and helping me to understand what exactly the data was and how it relates to the things I’m more used to.

A number of people have asked about printing the map. There is currently no formal plan to professionally print the map but maybe if we get enough enquiries or requests, you never know! The map sheet is quite large measuring 980 by 840mm, but if you have access to a large format printer or plotter then feel free to print off a copy for your bedroom or office wall but for any commercial use you must seek permission from OS as the map is copyrighted.
See creative commons license

I love the detail of the chaos’ with all the little stacks or plinths, with contours and shading around each at times it looks very busy but that is the landscape -- chaos
I love the detail of the chaos’ with all the little stacks or plinths, with contours and shading around each at times it looks very busy but that is the landscape – chaos

So onto the technicalities of the map creation process:

The first step was to consider the map layout. The purpose of good map layout is to allow a map to be read so there needs to be good flow across the map. It is good to try several different layouts and both GIS and vector graphics packages allow us to do this with relative ease. At OS, we have just updated our mid-scale paper map products with a new look legend and marginalia, so I used the template for this as the basis for my map layout but given my extent and legend information I opted to move the legend to be landscape across the bottom of the map sheet.

Initially I took an ArcGIS map document from Peter to study and understand the datasets thoroughly. Then I created a new blank document to start creating my version of the map, loading in the data to a scale and paper size that would suit my intended layout. At this point I reduced the size from a regular map sheet so that the map would fit more easily on display boards and to give me a nice round map scale of 1 to 4 million. We use ESRI ArcMap because it supports a number of planetary projections including the Mars 2000 projection I needed for this data.

Starting with the elevation data, I created an initial colour ramp for the base. This was quite difficult. Most printed maps, especially topographic maps from OS, have a white or no background colour. However a white map of a landscape devoid of any woods, bodies of water or manmade features makes for quite an empty map! So instead I considered where we have used layered colouring in the past either on published products such as the OS Tour Map and OS Wall Map series’ or on internal projects at larger scales. These coloured backdrops have retained an OS identity by the use of a soft, pastel colour palette that harmoniously sings OS with room left in the colour spectrum for the overlain features to do any shouting or solos.

Bands of colour didn’t really have the desired effect so I decided to show a continuous colour ramp based purely on the height of the land from the elevation data to tint the page.

Colour ramp before layer blending
Colour ramp before layer blending

Next I took the same elevation data and experimented with shaded relief, otherwise known as hillshading. Readers of my blog will know that I prefer to create hillshading in QGIS so this presented a dilemma. Do I keep the correct georeferencing in ArcGIS or lose that by manipulating the data in QGIS? After some experimentation I decided on the latter. QGIS will read in unrecognised geodata and assign a default ‘unknown’ projection.

In QGIS I create hillshading directly from the GeoTIFF elevation data (exactly as with OS Terrain data in asci grid format) by using the built-in hillshading tool – see previous post. To create good hillshading for printed maps one should really use a number of different lighting angles. I exported 10 different hillshadings from most directions but then combined these in Photoshop with a weighting or preference towards those lit from the North-Northwest (just like I usually would in Britain). The reason for including southern or eastern-based illuminations is to ensure you capture all of the detail of the terrain rather than lose some to extremes of light and shade.

Merging the hillshading in Photoshop is a repeatable process but my logic I’m afraid is too long-winded and unscientific to explain here. It is perhaps the ‘art’ in cartography. Especially when to capture the detail of the craters as well as the detail of the relatively flatter areas between (which are anything but flat by the way), I had to actually create two sets of hillshading at two different exaggerations. To highlight the crater detail I chose a 5x exaggeration and for the areas between I exaggerated by 50 times. So in effect I had 20 different hillshadings to merge in Photoshop and not all were to the same specification! Thankfully blending images in Photoshop is something I can do but to make life easier I merged each set of 10 separately and then combined the 2 final outputs, i.e. the 5x shading and the 50x shading.

Extracts of the final hillshading (before blending with other layers of the map)
Extracts of the final hillshading (before blending with other layers of the map)
We also have a mountain on the map but its diameter doesn’t afford it significant prominence. I have therefore tried to make it stand out with an Earthly snowy white peak.  Features like this benefit from a 3D visualisation.
We also have a mountain on the map but its diameter doesn’t afford it significant prominence. I have therefore tried to make it stand out with an Earthly snowy white peak. Features like this benefit from a 3D visualisation.

Back in ArcGIS, I next experimented with creating contours from the elevation data. Again this was no easy task. The data is incredibly detailed for such a vast, small-scale region and with such a ‘bumpy’ surface on Mars the resulting contours were incredibly jagged and ugly. Also the contour interval was difficult to get right for the same reason as the hillshading. The best interval for the craters was very different to the optimum interval for the areas between, but in this instance I was able through a long spell of trial and error to come up with a compromise contour interval of 250m. To solve the ugly contours I tried many different combinations of smoothing and simplification generalisation using the tools in ArcGIS and got some reasonable results but eventually I went back to the elevation data I had in QGIS and the tool in there gave more pleasing visual end results, so with hillshading already coming from QGIS I decided to use that for my contours too. To keep to the OS style brief I chose a brown-orange colour for the contours. I wanted them to be fairly thin and subtle but dark enough to be seen over the elevation ramp.

The one thing that any GIS seems to still struggle with is labelling contours. Producing nice ladders as we call them is not easy. There are many work arounds such as cutting ladders from the contour lines and then labelling those but with the close contour proximity in places I knew this would even be problematic so instead I decided to label the contours manually. A very painstaking, manually-intensive process but this is an area where map software really needs to improve. This is the one step that would need to be revisited for further map production.

I used Adobe Creative Suite for the contour labelling. I added the labels in Illustrator by loading the contour lines as a base, drawing smoothed curves alongside them and typing my labels as text on a line. Rather than use halos I decided to use Photoshop to create masks around my contour labels and then delete these selections from a rasterised set of the contour lines.

As well as the base colours, I did go back to ArcGIS to do an initial place of the nomenclature or place names and for the previous mission landing sites. Also in ArcGIS I placed some test traverse routes and other sample thematic overlay data which Peter had supplied for prototyping purposes.

An area of Martian Nomenclature
An area of Martian Nomenclature

As is my general preferred routine in paper map creation, I exported each layer from ArcGIS and QGIS as a high-resolution PDF with the exception of the base colours which I exported as raster and the hillshading which was already raster in Photoshop.

The next step, again as is commonplace in paper map creation, was to stack all of the layers in Illustrator to create the final map. I began by loading in the base map with the rough map layout and I was then able to bring in the unreferenced files (the hillshading and contours) and essentially drag them to fit the map window. Remember that I knew their extent was unchanged, all that had happened by sacrificing their georeferencing in QGIS was that they had become subtly warped in ratio or aspect due to misprojection, but as my initial experimentation had shown they were not too far out.

Above these I imported the grid, in cyan again to keep the OS theme, and labels. In ArcGIS I had placed all labels to be centred on top of their data point but once in Illustrator I moved each of them to their best cartographic position. Generally this is inside of a feature extent if it will fit otherwise it is around the feature with above right being a preferred location. That completed the map window or the body of the map.

Then I manually worked on improving the labelling of grid figures in the margin and on creating my legend working with the template I had as my starting point from OS Explorer Map.

Once complete I flatten the document and export as a CMYK PDF/X-4 file ready for print.

Ordnance Survey map of Mars
The final map sheet

A slight addendum to the above, I also created a simplified, reduced size version of the map to be more social-media friendly. This is still quite large but is legible if shrunk down to A3 size for desktop printing.

The map sheet also contains a locator map in the bottom right-hand corner. To make this I created a rectangular box around my extent as a shapefile, I downloaded some open data of the surface of Mars as a raster image, and I loaded them both into ArcGIS using the ‘Earth from Above’ projection. Basically I fooled ArcGIS into thinking that Mars was the Earth in order to get my location extent on a nice sphere and I customised the projection to centre on the centre of my extent… I hear the map projection nerds complaining already but it works!

Ordnance Survey map of Mars (reduced version)

Finally my colleagues and I have enjoyed the numerous comments on the map on websites and social media. Here are a few of our favourites:

“I’ve no idea why OS have bothered…
How about because they were asked? And also because they know a lot about maps. Just compared the image with Google Mars and the OS wins hands down on legibility and orientation and that’s because they’ve taken the trouble of not making something for computers that people may or may not be able to use.

And no doubt in response to The Register’s tagline ‘Plenty of craters, a bit short on pubs’:

They built a pub on mars once. It had to close, no-one would use it – no atmosphere!

See more of what I create along with my colleagues in the CartoDesign team at OS


Leave a Reply

Your email address will not be published. Required fields are marked *


Get every new post on this blog delivered to your Inbox.

Join other followers: