Map Design -- Rock Drawing

In a previous blog post (map design -- May 20, 2016), I mentioned the work that I had done in attempting to imitate the rock drawing style of Swiss topo maps. In the couple of years since then, I worked on a much larger project — a topographic map with rock drawing covering the entire area of the High Sierra. Several months after I had put that project aside, I realized that it might be worthwhile to write a few notes about how I went about creating the maps before it faded from my memory. So that's what this blog post is all about. I'm going to start with explaining how I made the maps shown at the end of my previous post, and then how I made the later High Sierra map.

First attempts at rock drawing

I chose to start with an attempt to map the area of the USGS 7.5’ Mount Ritter Quadrangle. I was familiar with the area and felt that it would be representative of much of the High Sierra terrain. I mostly just wanted to see what I could do with rock drawing, so my plan was to just overlay the drawing on the existing topo map. I first tried drawing lines in Adobe Illustrator following ridgelines and edges, thinking that some skeletal lines would be enough but it wasn't. Then I tried drawing in pencil on tracing paper. The result wasn’t too bad, but it was not easy. I couldn’t see the countours clearly through the tracing paper and the pencil tended to smudge. Attempts at erasing made a mess. After scanning, many areas of the drawing were too weak. I fit the drawings as an overlay on the topo map. I added a layer of shading derived from USGS dem data using software called SimpleDEMViewer. On the topo, the blue contour lines on the glaciers were much too dense. I replaced them with drawings done in Illustrator using the same contour interval as the rest of the map. It all turned out all right, but I thought that I could do better.

pencil on tracing paper overlaid on topo

I had been getting help from a pdf I found online entitled “Design Principles for Swiss-Style Rock Drawing” by Bernhard Jenny and others. I found that it pointed me in the right direction, but found it difficult to apply the style to real terrain in the High Sierra. The peaks and cliffs weren’t so hard to draw, but the lower slopes tended to look like random scratchings. From the countours it was hard to tell where rock drawing was appropriate. I tried using satellite photos and they helped a lot, but not entirely. Did moderately steep contours indicate rock slabs or talus slopes? I found that looking at satellite photos from multiple sources helped. In many places, the satellite photos on Google maps showed lots of snow cover which didn’t help in drawing the rocks. Photos from Terraserver from different times of day and different seasons often gave a better view.

I decided to try again, this time on a different area, the 15’ Matterhorn Peak quadrangle. I made a number of attempts at this, and after looking through my old files, I’m not sure exactly what I did each time. I know I used several different USGS topo maps as a base — the 1:100,000 Bridgeport quadrangle, the 1:62,500 Matterhorn Peak and the 1:24,000 Matterhorn Peak. I found the latter in two versions — one a jpeg image and a newer one that was all vector art. The vector art version was useful because I could separate the elements and use them in Adobe Illustrator, but I found that the contours were much less accurate than the older map. In some places like the cliffs of Yosemite Valley they were very inaccurate. But I used them anyway on some of my attempts because in most areas, their inaccuracy was not so obvious.

On a couple of maps I used the vector art contours but I wanted my finished map to be at a smaller scale so I changed the 40 ft. contours to 80 ft. contours by deleting every other line. I soon ran into problems because it was hard to tell where I had deleted and where I hadn’t. I solved that problem by changing the color of the lines rather than deleting them. When I was done, I could select all of the lines with changed color and delete them all at once. I also created contours from dem data using SimpleDEMViewer. The result was a bit-mapped image, but on some maps I converted it to vector art using the Live Trace feature in Illustrator.

I also found problems on all of the topo maps with the green indicating forest cover. On the bit-mapped topos it was difficult to separate the green color with Photoshop. I could easily separate the green areas from the vector art, but I didn’t think it was very accurately done. Comparing the maps with satellite photos, I saw that many of the green areas on the maps showed little or no trees in the photos. I spent a lot of time trying to re-draw the forest areas by tracing the photos.

My first step toward doing the rock drawing was to draw skeleton lines. On the topo map I traced all of the steep ridgelines and the tops and bases of cliffs, trying to simplify the terrain into a set of plane surfaces. At first I wondered if the skeletal lines alone would be enough, but soon found that they weren’t. The outlined surfaces needed to be filled with some kind of hachures to show the orientation. They needed to follow the direction of the light source, with lighter strokes for lighted surfaces and darker ones in shade.

For both the rock drawing and the shading I chose to use a light source from the west. It is more usual to use a light source from the northwest, but in the Sierra the major ridgelines tend to run northwest to southeast, and they don’t stand out as well with northwest light.

I tried several different approaches to the actual drawing. I did a couple of versions with line segments drawn in Adobe Illustrator. On one I used vertical lines for the steepest slopes and horizontal lines for the less steep, similar to the style of the Swiss topos. On another I used only vertical strokes, with dotted lines to indicate talus slopes. Both look pretty good, but because they are uniform line segments, they look rather stiff and mechanical. Then I tried drawing by hand with a fine-tipped pen, going back to the Swiss-style with both horizontal and vertical strokes. On tracing paper it was hard to see the map underneath. I tried using clear acetate, but it tended to smear and didn’t scan very well.

drawing in Illustrator -- all vertical strokes

drawn again with both vertical and horizontal strokes

Then I got the idea of printing the topo map with the image lightened by about 50% in photoshop, and drawing directly on the paper in black ink. After scanning, I used Photoshop curves to reduce the highlights by about 50% which left only the drawing showing. That worked quite well and I used that method to finish my 15’ Matterhorn Peak quadrangle and do another map at the same scale of Yosemite Valley. However, there were some problems with drawing freehand like that. One was that in order to get nice thin strokes, I needed to move my pen quickly, and my accuracy suffered. Another was that it was difficult to fix mistakes. I used white-out in places but it was messy to draw over it.

In trying to draw rocks, I soon found that contours were not enough to distinguish craggy rocks from talus or other steep slopes. I needed to study satellite photos to divide the map into areas for rock drawing, areas to indicate talus, and areas without rocks. For talus slopes, I tried using dot textures in Illustrator, but I think I ended up just stippling dots with my pen on the drawings.

On all of these maps, I needed to do a lot of image manipulation to get the various elements to fit. Elements derived from a single topo map were no problem, but the shading and contours from dem data and my rock drawings all had to be scaled, rotated and distorted in photoshop to get them into good registration.

section of final Matterhorn Peak map -- rock drawing in ink on paper

After I had finished the map of the Matterhorn Peak area, I decided to try using the same procedures to do one of Yosemite Valley. I felt like I had done pretty well on both of these maps, and put them aside for a year or so. Then I began to think again about the possibility of doing a map in a similar style of the whole High Sierra. I thought that if I used a smaller scale, like 1:100,000, I might be able to manage it. That scale would be just large enough to show the general shape of the mountains, but not a lot of fine detail, which would make the drawing easier. I intended the map to be viewed on screen rather than printed, so I would assemble it in Photoshop at 72 ppi which would keep the file at a manageable size (about 2 gb).

A map of the entire high sierra

I started by making a base map by fitting together 10 topos at the 1:100,000 scale. They were the quadrangles for Bridgeport, Excelsior Mts, Yosemite Valley, Benton Range, Shaver Lake, Bishop, Fresno, Mt Whitney, Visalia and Three Rivers. I deleted a number of 15’ quadrangle areas in the northeast and southwest that I didn’t want to bother with.

I wanted to start with the lines for lakes, streams and roads. I first considered downloading all of the 7.5’ vector art quadrangles, but realized that it would be a big job since I would need to download more than 200 files, each one would require a lot of work to extract the lines, and the lines would be much too detailed for the scale that I intended. Instead, I traced the lines from the 1:100,000 scale topos. By working in photoshop to lighten some of the image and use only the red channel, I got a gray-scale image that I could apply the auto-trace feature in Illustrator and get line work for a lot of the lakes and streams. I used a similar method on the green channel to trace the larger roads that were shown in red. In many places, my tracings were pretty good, with only a few gaps that needed to be connected and extra stuff to be deleted, but in other places, large areas had to be traced manually. Later, I went back and traced all of the trails, also done manually.

section of base map with water features, roads and trails

 Once again, I needed to map green areas to indicate forest. I felt that the green on topo maps was not accurate in many areas, and it varied from one map to another. Some of the quadrangles had no green at all. I remembered that for some previous maps I had used USGS land cover data. I pieced together screen captures from a viewer and got an image of the whole map area. After fitting it to my base map, I selected the green forest areas to use as a separate layer. Later, I decided that adding a very transparent layer of the whole land cover image improved the appearance of the map.

section of land cover map

My first step toward doing rock drawing was to do skeleton drawings for the whole area. I did these drawings as an overlay on the 15’ quadrangles that I intended to use for the final rock drawings. My plan was to do the drawings at the 1:62,500 scale and then reduce them to 1:100,000 scale on the final map. I drew lines following the contour lines for ridges and cliff edges, drawing only on areas with fairly steep contours. I refered to satellite images to distinguish steep talus from craggy rocks. This process took a long time to complete — probably several months. Looking back, I may not have needed to do these drawings as carefully as I did because when I did the actual rock drawing, I used the skeleton drawing mostly just to show where to draw rocks and where to leave blank.

section showing skeletal lines overlaid on topo map

Then I had to figure out how to do the rock drawing. I realized that drawing at a smaller scale than my previous efforts would require a little different style. I looked at examples of Swiss topos at a similar scale. I also looked at examples in Imhof’s book of smaller scale drawings. Some of them showed skeleton drawing in a nice looking style and I thought maybe that would work for me.

drawing using mostly skeletal lines

I planned to do the drawing in pen on paper like I did before. I printed sections of topo maps with skeleton drawings in blue and did test drawings in black ink on them. I tried imitating the Swiss style on some and only skeletion drawing on others. On these tests I ran into the same problems as I did on my larger scale maps. My pen strokes were not as accurate as I wanted and it was difficult to fix mistakes. Then I came up with the idea of drawing directly on the topo map in Photoshop. I could draw more precisely, I could easily see the underlying map image, and it was easy to erase and re-draw mistakes. I used the pencil tool set to a width of one pixel. Some of the lines needed to be thicker than others to indicate ridgelines and shading. To do that, I drew a single line and then traced over it slightly offset to make a thicker line. I could have reset the pen tool to a two pixel thickness, but that made it a bit too thick and mechanical looking. Double stroking the thicker lines was tedious but I felt that they gave me a better looking result. I decided not to use the skeleton drawing style that I had tried and went back to imitating the Swiss maps, using vertical strokes for cliff areas and horizontal strokes for less steep ones, with attention to light direction to add shading.

final drawing using Photoshop pencil tool

I started in the northern part of the map and worked south. It was a slow process, and when I had done maybe about a third of the area, I realized that my style had changed over time. It was looking much better than when I started. It was a bit discouraging, but I went back and re-worked many areas that I had already done in an attempt to create a uniform style. It took many months of work, but I finally finished drawing the whole map area. When the drawings were fitted to the 1:100,000 scale base map, they were reduced in size by 62.5% which softened the pixelation that showed at 72 dpi, and made the lines appear darker.

Next, I wanted to add representation of glaciers. The 15’ topos showed glaciers, but after comparing them with the 7.5’ topos, I saw that they were quite inconsistent and completely missing on some quadrangles. I ended up drawing glacier areas manually on the 15’ topos while looking at the 7.5’ topos on my iphone. I feel that my drawings are accurate enough since glaciers change from year to year anyway.

Shading and contours were done together using SimpleDEMviewer on DEM data downloaded from USGS. As before, I used shading with a light source from the west. Countours were set for 200 ft. Both shading and contours were saved as separate images, but kept together while resizing and fitting to the base maps to keep the two images in register.

All of the layers were fitted to the 1:100,000 topos, and then those were fitted to the large base map of the whole area. Quite a few areas appeared to be somewhat out of register, and they needed to be adjusted for a better fit. The transparency of the various layers was adjusted to try for the best final appearance.

Type for place names was added in Illustrator in two sections because the size of the map was too large to fit into a single file. The names were mostly copied from those on the base topo. Peak elevations were added mostly from the 15’ topos which were shown in feet, but in some cases converted from meters on the 7.5’ topos. The two series of topos don’t agree on many elevations, but I decided that it wasn’t worth worrying too much about.

section of finished map

When everything was together and looking good, I flattened the image and saved it as a jpeg, reducing a 1.83 gb file to 22.3 mb. The finished map is 146 in. x 227 in., rather big to print out on paper. Just to see what it would look like, I printed it out at half-size on 8.5x11 pages and spliced them together. It’s still rather awkward at 9.5 ft. long.

entire map area

As I was working on the map, I had in mind putting it online using some kind of software that allowed viewing the image at full size and easily scrolling from place to place. But when I was done, I was tired of it and just wanted to put it aside. I had spent a couple of years of obsessive work on it and needed to focus on other things. Now that half a year or so has passed, I can look at it again, but probably won’t do anything more for a while. Anyway, I wanted to write this explanation of what I did before it faded from memory.