Shaded relief from DEM in GRASS and QGIS

Question

I have been trying to get a good looking shaded relief model out of GRASS and Qgis for a day or so now and just can't seem to get it. The biggest problem I'm having is the valleys look like mountains, but I just can't seem to get the right combination in order to show mountains and valleys look right.

Bellow is a some pictures of the problems:

DEM (Brown is high and Red is low, red is only used for a visual check)

Shaded Model 1 with DEM Sun position 180 degress from North, sun height 80

Problem: Hills look good but the river (in red) still looks like a mountain even though it's the lowest point on the map.

Shaded Model 2 with DEM Sun position north west (315 degrees), sun height 60

Problem: Again the hills look good but the river (in red) still looks like a mountain even though it's the lowest point on the map.

I have tried all different combinations of angles and sun heights and I just can't seem to get the river to look like a river, it always looks like a hill.

Can anyone point me in the right direction on what I might be doing wrong?

EDIT: So I took the contour data home and ran the process from the start and used the defaults for r.shaded.relief and got this:

This looks much better to me, not the greatest colours but we can change that. I think my work install of grass must be busted because even the defaults looked bad.

Answer 1

For relief shading in GRASS, better use r.shaded.relief. It comes by default with light from west to east. Like this (SRTM example):

For my perception the valley looks ok in this example. The colortable comes from r.colors (there are "terrain" and "srtm", see here for examples of these color tables).

Answer 2

There's nothing wrong with this hillshading and the coloring certainly is not the problem. We are experiencing a common, well-known phenomenon: our visual processing inverts depth and height when the terrain appears to be illuminated from below. (Look at the image again while standing on your head: it will appear correct.) That's why it's best to place the light near the top (map "north"). A close look at these particular images suggests the sun directions are not what you think they are: the top one seems to be illuminated from the right, not the bottom; and the bottom one seems to be illuminated from the lower right, not the upper left. Adding or subtracting 180 degrees from either of the angles will solve the problem. If the results are usually surprising to you, suspect one or more of the following:

• The software might be using radians instead of degrees.

• The angle might be measured counterclockwise from due east (the math/physics convention).

Answer 3

On first glance, it seems like the color ramping used is what makes the data look particularly confusing, choosing an appropriate hypsometric tint would make the differences more obvious. You may also want to look at other pre-generated shaded relief datasets for comparison, such as this SRTM-based one.

Another thing that would likely help is to add a stream centerline in an obvious water color, this will make unambiguous which areas are low-lying on the map and prevent the optical trick which makes the river appear as a ridgeline.

You may want to also check out Tim Sutton's guide to making shaded relief DEMs using GDAL for some helpful advice and techniques using only command-line tools.

Answer 4

I agree with scw that part of the problem might be caused by the choice of colors. I think people tend to interpret green as "low" and red as "high". Compare:

http://www.icsm.gov.au/mapping/cartographic.html#tints

On sun positions:

It is based on the convention that an imaginary light source (eg the Sun) is in the northwest corner of the map and the shadows spread out towards the southeast corner.

Source: http://www.icsm.gov.au/mapping/cartographic.html#shading