Relief Visualization Toolbox, ver. 1.

3
Manual
By Žiga Kokalj, Klemen Zakšek, Krištof Oštir, Peter Pehani and Klemen Čotar
Research Centre of the Slovenian Academy of Sciences and Arts
Contact: [email protected]
When using the toolbox, please cite:
Kokalj, Žiga, Klemen Zakšek and Krištof Oštir. 2011. Application of Sky-View Factor for the Visualization of
Historic Landscape Features in Lidar-Derived Relief Models. Antiquity 85 (327): 263–273.
Zakšek, Klemen, Krištof Oštir and Žiga Kokalj. 2011. Sky-View Factor as a Relief Visualization Technique.
Remote Sensing 3: 398–415.

General information
This software was produced to help scientists visualize raster elevation model datasets. We have
narrowed down the selection to include techniques that have proven to be effective for the identification
of small scale features. Default settings therefore assume working with high resolution digital elevation
models, derived from airborne laser scanning missions (lidar). Despite this, the techniques can also be
used for different other purposes. Sky-view factor, for example, can be efficiently used in numerous
studies where digital elevation model visualizations and automatic feature extraction techniques are
indispensable, e.g. in geography, geomorphology, cartography, hydrology, glaciology, forestry and
disaster management. It can be used even in engineering applications, such as predicting the availability
of the GPS signal in urban areas.
For a more detailed description of the visualization methods see the references given at each method,
and a comparative paper describing them (e.g. Kokalj et al. 2013).
The tool also supports elevation raster file data conversion. It is possible to convert all frequently used
single band raster formats into GeoTIFF, ASCII gridded XYZ, Erdas Imagine file and ENVI file formats.

RVT tool GUI for computation of different visualizations (left) and a single band raster data converter
(right).

1
Visualization techniques showing St. Helena church and its immediate surroundings. St. Helena is a known
yet un-researched archaeological site west of Kobarid, Slovenia, believed to be a late Roman camp.

Online resource
http://iaps.zrc-sazu.si/en/rvt
Check for updates from time to time. Please report any bugs and suggestions for improvements.

2
Version
Version: 1.3, September 2016
Name of the standalone package: RVT_1.3_Win64.zip (works independently).
For changes see version history at the bottom.

Installation
No installation is required. Unzip the package RVT_1.3 to any folder and run the exe file.

Input and output files

Input file(s): one or several digital elevation model file(s) in GeoTIFF format or any GDAL (GDAL
Development Team 2014) supported format (e.g. GeoTIFF, generic binary file, Erdas Imagine file, ENVI file,
Arc/Info ASCII Grid, ASCII gridded XYZ, JPEG2000…). If your extension is not listed on the Add files menu,
change the format filter to *.*. Input files can come from multiple folders and can be of different formats.
You can copy-paste the file list into the input window or manually type in the files. Each path/filename
has to be in a separate line. For ASCII gridded XYZ input files the software assumes that units are meters
and that coordinates have even spacing, therefore, it will not convert ungridded XYZ data, e.g. last return
lidar data.
Output file formats for data format conversion: GeoTIFF, ASCII gridded XYZ, Erdas Imagine file or ENVI file.
Output files for visualizations: a pair of GeoTIFFs per each selected visualization:
- a calculated 32-bit result, and
- a simplified 8-bit result, optimized for non-GIS software.
All output files are written into the folder of the input file. Output file names for visualizations are
composed of the input file name, and suffixes describing the selected method and processing parameters.
Format conversion only changes the file extension. N.B. If output files already exist, the tool replaces
them without warning! It is possible to disable this if you uncheck the option.
Each execution of the program generates a processing log file per input file that includes a list of
performed visualization methods and parameters used, output file names, possible warnings, and other
metadata. The log file is named input_file_name_process_log_yyyy-mm-dd-hh-mm-ss.txt.
Simplified 8-bit GeoTIFF files are prepared for displaying the results in non-GIS software, e.g. by Windows
Photo Viewer or by Preview for Mac users. Each 8-bit visualization uses its own histogram stretch, as
described in the table below. The histogram stretch with a cut-off does not work when there are more
than 2% extreme values such as no-data values or outliers (e.g. 0 value borders, “birds”, “clouds”…) – the
8-bit image is grey. You can visualize the results by applying a manual stretch (about -0.1 to 0.1 m for
SLRM for example) to the original results.
Name suffix Histogram Min, max Note
stretch type
Analytical hillshading HS linear stretch 0, 1
Hillshading from MULTI-HS linear stretch 0, 1 Red 315°, Green 15°, Blue 75°.
multiple directions
PCA of hillshading PCA histogram 2% cut-off
equalization
Slope gradient SLOPE linear stretch 0°, 51° inverted greyscale bar
Simple local relief model SLRM histogram 2% cut-off
equalization
Sky-view factor SVF linear stretch 0.6375, 1

3
 Anisotropic sky-view SVF-A histogram 2% cut-off
factor equalization
Openness – positive OPEN-POS linear stretch 60°, 95°
Openness – negative OPEN-NEG linear stretch 60°, 95°
Sky illumination model SIM linear stretch 0.25% cut-off
at minimum
Local dominance LD linear stretch 0.5, 1.8

As with all spatial calculations that consider neighbourhood, edge pixels do not have correct values. The
size of the incorrect edge depends on the neighbourhood size; e.g. when openness is calculated with
search radius of 10 pixels, 10 edge pixels have incorrect values.

Methods and parameters

To ease the usage of the toolbox the number of input parameters required for each visualization
technique is kept to a minimum.
If you select a parameter beyond the allowed interval (min … max), the parameter is adjusted (trimmed)
to fit into the interval, and a corresponding warning is written in the log text file.
Vertical exaggeration is the common parameter that influences all techniques. You can set it higher than
the default 1, if you need more contrast in the results; e.g. set it to 3 if the terrain is very flat, or 20 if you
use very detailed models, derived with structure-from-motion.
Default Allowed values Most useful Note
value min … max values
Vertical exaggeration 1 -1000 … 1000 0.5 … 3 No exaggeration = 1.
factor For flat relief use > 1.
Use < 1 and > 0 if your terrain data has been
converted to integer (whole) values (e.g. use
0.001 if units are mm).
For calculations on inverted relief use
negative values.

Analytical hill-shading is straightforward to interpret even by non-experts and without training. However
direct illumination restricts the visualization in dark shades and brightly lit areas, where no or very little
detail can be perceived. A single light beam also fails to reveal linear structures that lie parallel to it which
can be problematic in some applications, especially in archaeology.
Default Allowed values Most useful Note
value min … max values
Sun azimuth [°] 315 0 … 360 0 … 90, 0 is North and 90 is East. Values from
270 … 360 southern hemisphere (90 … 270) display
inverted shaded relief.
Sun elevation angle [°] 35 0 … 90 5 … 45 Use small values (5, 10) for flat terrain and
higher values (45) for steep terrain.

Analytical hillshading can be calculated in multiple directions that are equally distributed between 0° and
360°. 0° is always in band 1, followed by azimuths in clockwise direction, e.g. 45° in band 2, 90° in band 3
… 315° in band 8, for calculation in 8 directions. The 8-bit image is a result of calculation in three
directions, separated by 60° (315° in the red band, 15° in the green band, 75° in the blue band).
Default Allowed values Most useful Note
value min … max values

4
 Number of directions 16 4 …360 8, 16, 32, 36 The drop down menu is editable, so any
number in the allowed range can be
inserted. However, it is not very useful to go
much beyond 16 directions, due to high
autocorrelation.
Sun elevation angle [°] 35 0 … 75 5 … 45 Use small values (5, 10) for flat terrain and
higher values (45) for steep terrain.

Principal Components Analysis (PCA) is a mathematical procedure that summarizes the information of
correlated data; hillshaded images from multiple directions in this case. The method does not provide
consistent results with different datasets. Some common examples of displaying the components are:
- a combination of the first and second principal components, transparently overlaid,
- a false colour composite image (RGB) of the first three components, or
- displaying the third component on its own with high histogram stretch and clipping.
The 8-bit image shows the first three components as an RGB image (1st component in the red band, 2nd in
the green band, 3rd in the blue band).
Other parameters are set at the hillshading from multiple directions method box (see above).
Default Allowed values Most useful Note
value min … max value
Number of components 3 1 … (number of 3 The first three components usually hold
to save directions-1) more than 99% of the information of the
original hillshaded images. The rest is
“noise”. But sometimes you are interested in
it, because it presents something
uncommon.

Slope gradient represents the maximum rate of change between each cell and its neighbours and can be
calculated either as degree of slope (as in this tool) or as percentage of slope. If presented in an inverted
greyscale (steep slopes are darker), slope severity retains a very plastic representation of morphology.
However, additional information is needed to distinguish between positive/convex (e.g. banks) and
negative/concave (e.g. ditches) features since slopes of the same gradient (regardless of rising or falling)
are presented with the same colour. The method requires no parameters.
Local relief modelling removes the large scale morphological elements (hills, valleys…) from data so only
small scale features remain (e.g. archaeology). This version of the tool uses a simplified process – the
trend is computed by a simple mean filter and a trend removed model is produced directly by subtracting
the filtered model from the original. For a more complex method see Hesse (2010) and LiVT (Hesse 2013).
Default Allowed values Most useful Note
value min … max values
Radius for trend 20 5 … 50 10 … 50 Radius should be a bit more than half the
assessment [px] size of the features you are interested in.

Sky-view factor is a proxy for diffuse illumination and measures the proportion of the sky visible from a
given point. Locally flat terrain, ridges and earthworks (e.g. building walls, cultivation ridges, burial
mounds) which receive more illumination are highlighted and appear in light to white colours on a SVF
image, while depressions (e.g. trenches, moats, ploughing furrows, mining pits) are dark because they
receive less illumination (Zakšek et al. 2011).
The option to remove noise does not consider nearest pixels in the calculation. This diminishes small
variations that are usually a result of data collection and processing, and are seen as “salt and pepper
effect” on a sky-view factor or openness image.

5
 Default Allowed values Most useful Note
value min … max values
Number of search 8 4 … 360 8 … 36 Computational time increases as a linear
directions function of number of search directions. 16 is
optimal for most applications. The drop
down menu is editable, so any number in the
allowed range can be inserted.
Search radius [px] 10 1 … 100 5 … 50 Use small search radius (5-10 m; i.e. 5-20
pixels) if you are interested in small features,
e.g. archaeology.
Remove noise none low … high low low: first 10% pixels (1 pixel if r <= 10)
medium: first 20% pixels (2 pixels if 5 < r
<=10)
high: first 40% pixels (4 pixels if r <=10)

Anisotropic sky-view factor assumes that the sky is brighter in some directions than in others. The
weights are based on the cosine function of half angle (Zakšek et al. 2012). There are three parameters
that typically have to be set – the azimuth of the highest weight, the exponent defining the gradient from
maximal to minimal weight, and the minimal possible weight. The exponent and minimal weight define
the level of anisotropy – the greater the exponent and the smaller the minimal weight – the stronger the
effect. This has been simplified to the low and high anisotropy levels where the minimal possible weights
are 0.4 and 0.1, and the exponents are 4 and 8 respectively.
Other parameters are set at the sky-view factor method box (see above).
Default Allowed values Most useful Note
value min … max value
Anisotropy level low low, high low “How much the relief perception is
enhanced.”
Main direction of 315 0 … 360 315 0 is North and 90 is East. 315 presents usual
anisotropy [° ] direction for shaded relief.

Openness is also a proxy for diffuse illumination and is based on an estimation of a mean horizon
elevation angle within a defined search radius. The mean value of all zenith angles gives positive
openness, while the mean nadir value gives negative openness. Positive openness is similar to sky-view
factor, with a more “flattened feel”, while negative openness gives additional information on convex
features (Yokoyama et al. 2002). Because it is direction and shading independent and removes general
topography, it is useful for automatic feature recognition.
Parameters are set at sky-view factor method box (see above).
Sky illumination models quantitatively represent natural luminance of the sky under various atmospheric
conditions (Kennelly and Stewart 2014). Uniform and overcast sky models are implemented as they both
disregard directional shadowing effects. Mode details in shadows can be seen using the overcast sky
model. Calculations last much longer than for other visualizations, especially with large maximum shadow
modelling distances.
Default Allowed values Most useful Note
value min … max values
Sky model overcast overcast, overcast Different weights are assigned to calculation
uniform points. All weights are equal for uniform sky
model while weights increase towards the
zenith for the overcast model.
Number of sampling 250 250 or 500 250 Defines how smooth is the resulting image.
points The difference is minimal.

6
 Maximum shadow 100 5 … unlimited 100 … 500 The maximum distance to which shadows of
modelling distance [px] objects are calculated.

Local dominance visualisation of a DEM shows how dominant an observer standing above a certain pixel
would be for a local surrounding area (Hesse 2016). Dominance is higher for points on local protrusions as
well as on slopes, and lower for points in local depressions. It is well suited for very subtle positive relief
features such as former field boundaries or heavily eroded burial mounds, but also delivers very good
results for topographic depressions such as dolines, mining traces, and hollow ways.
Default Allowed values Most useful Note
value min … max values
Minimum radius [px] 10 5 … 20 Minimum radius should be a bit more than
half the size of the features you are
interested in.
Maximum radius [px] 20 10 … 50 Maximum radius should be a bit more than
the size of the features you are interested in.

Mosaicking
RVT supports mosaicking of multiple files, e.g. tiled lidar elevation models. The method is simple and not
many checks are performed, but works well if data are reasonable, i.e. if all the files have the same
projection, resolution, square pixels etc. The result is written in the same location and has the same name
as the first input file with a _mosaic suffix. All the input files are converted to GeoTIFFs.

Settings file
Default settings can be changed in the default_settings.txt file in the RVT_1.3_Win64\settings directory.
Computation of a visualization is turned on if its name is followed by 1 and off if it is followed by 0, e.g.
hillshading = 1, multiple_hillshading = 0. For the meaning of other settings see descriptions and tables at
individual visualizations above.

Processing without a GUI

RVT checks for filenames with paths in the process_files.txt file in the RVT_1.3_Win64\settings directory.
If any filenames are found, the GUI does not show, and all the files are processed with the settings from
the settings file. The filenames should be listed in rows; see below for example.
D:\Podatki\Lidar\Rence\rence_DEM_05m.tif
D:\Podatki\Lidar\Germany\Hugelsheim_DTM_1m.asc
D:\Documents\Papers\2016 Lidar visualization guidelines\data\Maiden_castle_DSM_05m.tif
D:\Documents\Papers\2016 Lidar visualization guidelines\data\Odolina_dem_1m.tif
C:\Test\Manhattan_DSM_1m_avg.xyz

License agreement
This software is distributed without any warranty and without even the implied warranty of
merchantability or fitness for a particular purpose.

Known issues
IDL running on Windows can encounter problems with large arrays (depending on the amount of physical
RAM, limitations of the operating system and memory fragmentation; see documentation on Exelis VIS
web pages). To help overcome this issue the SVF and openness based methods use tiling. The tiles are

7
maximum 5 megapixels large (e.g. 2500 by 2000 pixels). A buffer size is determined from selected
visualization parameters.
The software can crash when processing large DEMs on a lot of shading directions as a result of a
limitation of GeoTIFF file size.
RVT can only process single band files.

Version history
Version 1.3, September 2016
- Added support for automatic processing without GUI.
- Local dominance modelling.
Version 1.2, August 2015
- Added support for overcast and uniform sky illumination modelling.
- Settings can be read from a txt file and last used settings are saved temporarily.
- It is now possible to mosaic multiple files.
- SLRM calculation was changed from Gaussian averaging to a simple mean.
Version: 1.1, October 2014
- Added support for processing of multiple files.
- The tool now reads different file formats and can convert them to GeoTIFF, ASCII gridded XYZ,
Erdas Imagine file or ENVI file.
Version: 1.0, November 2013
- Nine relief visualisation methods implemented: hillshading, hillshading from multiple directions,
PCA of hillshading, slope gradient, simple local relief model, sky-view factor, anisotropic sky-view
factor, positive and negative openness implemented.

References
GDAL Development Team. 2014. GDAL - Geospatial Data Abstraction Library. Windows (version 1.11.0).
Open Source Geospatial Foundation.
Hesse, Ralf. 2010. LiDAR-Derived Local Relief Models - a New Tool for Archaeological Prospection.
Archaeological Prospection 17 (2): 67–72.
———. 2013. Lidar Visualization Toolbox. Visual Basic .NET (version 1.0.0.20).
http://sourceforge.net/projects/livt/?source=navbar.
———. 2016. Visualisierung Hochauflösender Digitaler Geländemodelle Mit LiVT. In
Computeranwendungen Und Quantitative Methoden in Der Archäologie. 4. Workshop Der AG CAA
2013, edited by U. Lieberwirth and I. Herzog, Edition Topoi, 109–128. Berlin Studies of the
Ancient World. Berlin: Topoi.
Kennelly, Patrick J. and A. James Stewart. 2014. General Sky Models for Illuminating Terrains.
International Journal of Geographical Information Science 28 (2) (February 1): 383–406.
doi:10.1080/13658816.2013.848985.
Kokalj, Žiga, Klemen Zakšek and Krištof Oštir. 2013. Visualizations of Lidar Derived Relief Models. In
Interpreting Archaeological Topography – Airborne Laser Scanning, Aerial Photographs and
Ground Observation, edited by Rachel Opitz and C. David Cowley, 100–114. Oxford: Oxbow
Books.
Yokoyama, Ryuzo, Mlchlo Shlrasawa and Richard J. Pike. 2002. Visualizing Topography by Openness: A
New Application of Image Processing to Digital Elevation Models. Photogrammetric Engineering
and Remote Sensing 68: 251–266.
Zakšek, Klemen, Krištof Oštir and Žiga Kokalj. 2011. Sky-View Factor as a Relief Visualization Technique.
Remote Sensing 3: 398–415.

8
Zakšek, Klemen, Krištof Oštir, Peter Pehani, Žiga Kokalj and Ekkehard Polert. 2012. Hill Shading Based on
Anisotropic Diffuse Illumination. In Symposium GIS Ostrava 2012, 1–10. Ostrava: Technical
University of Ostrava.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Copyright © 2016 Research Centre of the Slovenian Academy of Sciences and Arts (ZRC SAZU)
16/9/2016