Sine wave, clustering and watershed analysis to implement adaptive illumination and generalization in shaded relief representations
ISBN 978-85-88783-11-9
Authors
1Serebryakova, M.; 2Veronesi, F.; 3Hurni, L.
1ETH ZURICH Email: mserebry@ethz.ch
2ETH ZURICH Email: fveronesi@ethz.ch
3ETH ZURICH Email: lhurni@ethz.ch
Abstract
Shaded relief representations were traditionally produced manually by specifically trained cartographers. This was a labour-intensive and time-consuming task, but even today its results are considered superior to analytical hillshadings, in particular for mountainous landscapes. However, manual shading was to a significant extent a subjective process and its lack of standardization is the reason why many authors have developed new methods to increase the visual quality of shaded relief representations, by automating the guidelines outlined e.g. by Imhof. Their efforts elevated the visual quality of analytical hillshading, but had only limited success in creating automatic and standardized methods and workflows for re-creating specific cartographic terrain representation techniques . In the last two years, in the Institute of Cartography and Geoinformation of ETH Zurich, we worked on creating GIS tools to increase the visual quality of shaded relief representations. The major development of the project was the creation of a method, based on a sine wave equation, to continuously change the orientation of the light vector. This method was originally developed to work independently by changing the light direction based only upon aspect. By solving the sine wave equation using aspect, we are able to seamlessly change the light direction between two light sources, thus producing a much more informative shaded relief. Since this method is purely based on aspect and a mathematical equation, it can be applied with consistency, and being available as an ArcGIS toolbox, it is also extremely easy to use even by non-cartographers. This method however does not support any form of terrain generalization, which is very useful because on the one hand it allows a cartographer to hide distracting details and on the other hand provides a way to highlight important geomorphologic features. For this reason we implemented two additional methods, based on clustering aspect and watershed analysis, to overcome this limitation. These two methods are very different in nature, clustering aspect is fully automatic but requires some tuning to eliminate noise and smooth the clustered areas. Only after these corrections we can obtain clustered areas that are geomorphologically meaningful. However, even then not all the relevant areas are properly covered, in some cases smoothed clusters do not stop at ridges and this causes some artefacts. On the contrary, watershed analysis allows us to precisely extract areas between ridges and drainage, which is the general focus of shaded relief. However, the process is not automatic, it requires user intervention and its results depend upon the user experience and the accuracy of the algorithm. Finally, the use of multiple light sources is not the only visual device used in manual shading. Changes of tone are also widespread to increase the contrast differences between valley and ridges and as a consequence highlight all the most important geomorphological shapes. To automate this aspect we developed three possible solutions. The first two methods transform elevation or slope into weights and then modify the light vector inclination based on them. This allows us to set the inclination of the light vector to 90° for low elevation (i.e. valley) in order to obtain a constant whitish tone that masks urban areas and the noise in the DTM along highly populated valley. The inclination then decreases with elevation up to a minimum value, which can be set by the user, to gradually increase the image tone along slopes. The final approach we developed was the translation of the methods suggested by Brassel and Jenny into ArcGIS. These two methods allow us within the ArcGIS platform to simulate both tonal changes and the aerial perspective, which are essential parts of the Swiss style of shading. The corrections guarantee a blackish tone along sharp and high ridges (on the shadow side) and white on the light side as Imhof suggested.