| 1 |
<h2>DESCRIPTION</h2> |
| 2 |
|
| 3 |
<b>r.horizon</b> computes the angular height of terrain horizon in |
| 4 |
radians. It reads a raster of elevation data and outputs the horizon |
| 5 |
outline in one of two modes: |
| 6 |
|
| 7 |
<ul> |
| 8 |
<li> point: as a series of horizon |
| 9 |
heights in the specified directions from the given point(s). The results are |
| 10 |
written to the stdout.</li> |
| 11 |
<li> raster: in this case the output is |
| 12 |
one or more raster maps, with each point in a raster giving the horizon |
| 13 |
height in a specific direction. One raster is created for each direction.</li> |
| 14 |
</ul> |
| 15 |
|
| 16 |
<p> |
| 17 |
The directions are given as azimuthal angles (in degrees), with |
| 18 |
the angle starting with 0 towards East and moving counterclockwise |
| 19 |
(North is 90, etc.). The calculation takes into account the actual |
| 20 |
projection, so the angles are corrected for direction distortions |
| 21 |
imposed by it. The directions are thus aligned to those of the |
| 22 |
geographic projection and not the coordinate system given by the rows |
| 23 |
and columns of the raster map. This correction implies that the |
| 24 |
resulting cardinal directions represent true orientation towards the |
| 25 |
East, North, West and South. The only exception of this feature is |
| 26 |
x,y coordinate system, where this correction is |
| 27 |
not applied. |
| 28 |
|
| 29 |
<p> |
| 30 |
Using the <b>-c</b> flag, the azimuthal angles will be printed in compass |
| 31 |
orientation (North=0, clockwise). |
| 32 |
|
| 33 |
<p> |
| 34 |
Activating the <b>-l</b> flag allows additionally printing the distance |
| 35 |
to each horizon angle. |
| 36 |
|
| 37 |
<h3>Input parameters:</h3> |
| 38 |
<p>The <i>elevation</i> parameter is an input elevation raster map. If |
| 39 |
the buffer options are used (see below), this raster should extend |
| 40 |
over the area that accommodate the presently defined region plus |
| 41 |
defined buffer zones. |
| 42 |
|
| 43 |
<p>The <i>step</i> parameter gives the angle step (in degrees) |
| 44 |
between successive azimuthal directions for the calculation of the |
| 45 |
horizon. Thus, a value of 5 for the <i>step</i> will give a total of |
| 46 |
360/5=72 directions (72 raster maps if used in the raster map mode). |
| 47 |
|
| 48 |
<p>The <i>start</i> parameter gives the angle start (in degrees) |
| 49 |
for the calculation of the horizon. The default value is 0 (East with |
| 50 |
North being 90 etc.). |
| 51 |
|
| 52 |
<p>The <i>end</i> parameter gives the angle end (in degrees) |
| 53 |
for the calculation of the horizon. The end point is omitted! |
| 54 |
So for example if we run r.horizon with step=10, start=30 and end=70 |
| 55 |
the raster maps generated by r.horizon will be only for angles: |
| 56 |
30, 40, 50, 60. The default value is 360. |
| 57 |
|
| 58 |
<p>The <i>direction</i> parameter gives the initial direction of the |
| 59 |
first output. This parameter acts as an direction angle offset. For |
| 60 |
example, if you want to get horizon angles for directions 45 and 225 |
| 61 |
degrees, the <i>direction</i> should be set to 45 and <i>step</i> to |
| 62 |
180. If you only want one single direction, use this parameter to |
| 63 |
specify desired direction of horizon angle, and set the <i>step</i> |
| 64 |
size to 0 degrees. Otherwise all angles for a given starting <i>direction</i> |
| 65 |
with step of <i>step</i> are calculated. |
| 66 |
|
| 67 |
<p>The <i>distance</i> controls the sampling distance step size for the |
| 68 |
search for horizon along the line of sight. The default value is 1.0 |
| 69 |
meaning that the step size will be taken from the raster resolution. |
| 70 |
Setting the value below 1.0 might slightly improve results for |
| 71 |
directions apart from the cardinal ones, but increasing the |
| 72 |
processing load of the search algorithm. |
| 73 |
|
| 74 |
<p>The <i>maxdistance</i> value gives a maximum distance to move away |
| 75 |
from the origin along the line of sight in order to search for the |
| 76 |
horizon height. |
| 77 |
<!-- check next statement in the source code --> |
| 78 |
The default <i>maxdistance</i> is the full map extent. |
| 79 |
|
| 80 |
The smaller this value the faster the calculation but |
| 81 |
the higher the risk that you may miss a terrain feature that can |
| 82 |
contribute significantly to the horizon outline. Note that a viewshed |
| 83 |
can be calculated with <em>r.viewshed</em>. |
| 84 |
|
| 85 |
<p>The <i>coordinate</i> parameter takes one or |
| 86 |
multiple pairs of easting-northing values |
| 87 |
in the current coordinate system and calculates the values of angular |
| 88 |
height of the horizon around each point. To achieve the |
| 89 |
consistency of the results, the point coordinate is |
| 90 |
aligned to the midpoint of the closest elevation raster cell. |
| 91 |
|
| 92 |
<p>If an analyzed point (or raster cell) lies close to the edge of |
| 93 |
the defined region, the horizon calculation may not be realistic, |
| 94 |
since it may not see some significant terrain features which could |
| 95 |
have contributed to the horizon, because these features are outside |
| 96 |
the region. There are to options how to set the size of the buffer |
| 97 |
that is used to increase the area of the horizon analysis. The |
| 98 |
<i>bufferzone</i> parameter allows you to specify the same size of |
| 99 |
buffer for all cardinal directions and the parameters <i>e_buff</i>, |
| 100 |
<i>n_buff</i>, <i>s_buff</i>, and <i>w_buff</i> allow you to specify |
| 101 |
a buffer size individually for each of the four directions. The |
| 102 |
buffer parameters influence only size of the read elevation map, |
| 103 |
while the analysis in the raster mode will be done only for the |
| 104 |
area specified by the current region definition. |
| 105 |
|
| 106 |
<p>The <i>output</i> parameter defines the basename of the output |
| 107 |
horizon raster maps. The raster name of each horizon direction |
| 108 |
raster will be constructed as <i>basename_</i>ANGLE, where ANGLE |
| 109 |
is the angle in degrees with the direction. If you use <b>r.horizon</b> |
| 110 |
in the point mode this option will be ignored. |
| 111 |
|
| 112 |
<p>The <i>file</i> parameter allows saving the resulting horizon |
| 113 |
angles in a comma separated ASCII file with option <b>format=plain</b> |
| 114 |
or in a JSON file with option <b>format=json</b> (point mode only). If |
| 115 |
you use <b>r.horizon</b> in the raster map mode this option will be ignored. |
| 116 |
|
| 117 |
<p>At the moment the elevation and maximum distance must be measured in meters, |
| 118 |
even if you use geographical coordinates (longitude/latitude). If your |
| 119 |
projection is based on distance (easting and northing), these too must |
| 120 |
be in meters. The buffer parameters must be in the same units as the |
| 121 |
raster coordinates (e.g., for latitude-longitude buffers are |
| 122 |
measured in degrees). |
| 123 |
|
| 124 |
<h2>NOTES</h2> |
| 125 |
|
| 126 |
<h3>Method</h3> |
| 127 |
|
| 128 |
The calculation method is based on the method used in <b>r.sun</b> |
| 129 |
to calculate shadows. It starts at a very shallow angle and walks |
| 130 |
along the line of sight and asks at each step whether the line of |
| 131 |
sight "hits" the terrain. If so, the angle is increased to |
| 132 |
allow the line of sight to pass just above the terrain at that point. |
| 133 |
This is continued until the line of sight reaches a height that is |
| 134 |
higher than any point in the region or until it reaches the border of |
| 135 |
the region (see also the <i>bufferzone,e_buff</i>, <i>n_buff</i>, |
| 136 |
<i>s_buff</i>, and <i>w_buff</i>). The number of lines of sight (azimuth |
| 137 |
directions) is determined from the <i>direction</i> and |
| 138 |
<i>step</i> parameters. The method takes into account the curvature |
| 139 |
of the Earth whereby remote features will seem to be lower than they |
| 140 |
actually are. It also accounts for the changes of angles towards |
| 141 |
cardinal directions caused by the projection (see above). |
| 142 |
|
| 143 |
<p> |
| 144 |
The output with the <b>-d</b> flag is azimuth degree (-90 to 90, where |
| 145 |
0 is parallel with the focal cell). |
| 146 |
|
| 147 |
<p> All horizon values are positive (or zero). While negative values are |
| 148 |
in theory possible, <b>r.horizon</b> currently does not support them. |
| 149 |
|
| 150 |
|
| 151 |
<h3>Performance</h3> |
| 152 |
Parallel processing is implemented for the raster mode. |
| 153 |
To enable parallel processing, the user can specify the number of threads |
| 154 |
to be used with the <b>nprocs</b> parameter. |
| 155 |
Figures below show benchmark results running on |
| 156 |
Intel® Core™ i5-13600K CPU @ 3.5GHz. |
| 157 |
See benchmark scripts |
| 158 |
in the source code for more details. |
| 159 |
|
| 160 |
<p> |
| 161 |
The iterations of the algorithm used in <b>r.horizon</b> |
| 162 |
depends on the topography. As a result, the benchmark results may vary |
| 163 |
depending on the topography of the study area. |
| 164 |
|
| 165 |
<div align="center" style="margin: 10px"> |
| 166 |
<img src="rhorizon_raster_time.png" |
| 167 |
alt="time for r.horizon with different map sizes" border="0"> |
| 168 |
<img src="rhorizon_raster_speedup.png" |
| 169 |
alt="speedup for r.horizon with different map sizes" border="0"> |
| 170 |
<img src="rhorizon_raster_efficiency.png" |
| 171 |
alt="efficiency for r.horizon with different map sizes" border="0"> |
| 172 |
<br> |
| 173 |
<i>Figure: Benchmark shows execution time, parallel speedup and |
| 174 |
efficiency for different numbers of cells (1M, 2M, 4M, and 8M).</i> |
| 175 |
</div> |
| 176 |
|
| 177 |
|
| 178 |
<h2>EXAMPLES</h2> |
| 179 |
|
| 180 |
The examples are intended for the North Carolina sample dataset. |
| 181 |
|
| 182 |
<h3>Point mode</h3> |
| 183 |
|
| 184 |
<b>Example 1</b>: determine horizon angle in 225 degree direction (output |
| 185 |
of horizon angles CCW from East): |
| 186 |
|
| 187 |
<div class="code"><pre> |
| 188 |
g.region raster=elevation -p |
| 189 |
r.horizon elevation=elevation direction=215 step=0 bufferzone=200 \ |
| 190 |
coordinates=638871.6,223384.4 maxdistance=5000 |
| 191 |
</pre></div> |
| 192 |
<p> |
| 193 |
<b>Example 2</b>: determine horizon values starting at 90 deg (North), |
| 194 |
step size of 5 deg, saving result as CSV file: |
| 195 |
|
| 196 |
<div class="code"><pre> |
| 197 |
r.horizon elevation=elevation direction=90 step=5 bufferzone=200 \ |
| 198 |
coordinates=638871.6,223384.4 maxdistance=5000 file=horizon.csv |
| 199 |
</pre></div> |
| 200 |
|
| 201 |
<p> |
| 202 |
<b>Example 3</b>: test point near highway intersection, saving result |
| 203 |
as CSV file for plotting the horizon around the highway intersection: |
| 204 |
|
| 205 |
<div class="code"><pre> |
| 206 |
g.region n=223540 s=220820 w=634650 e=638780 res=10 -p |
| 207 |
r.horizon elevation=elevation direction=0 step=5 bufferzone=200 \ |
| 208 |
coordinates=636483.54,222176.25 maxdistance=5000 -d file=horizon.csv |
| 209 |
</pre></div> |
| 210 |
|
| 211 |
<center> |
| 212 |
<img src="rhorizon_shaded_dem_point.png"><br> |
| 213 |
Test point near high way intersection (North Carolina sample dataset) |
| 214 |
<p> |
| 215 |
<img src="rhorizon_singlepoint_plot.png"><br> |
| 216 |
Horizon angles for test point (CCW from East) |
| 217 |
</center> |
| 218 |
|
| 219 |
<p>We can plot horizon in polar coordinates using Matplotlib in Python: |
| 220 |
<div class="code"><pre> |
| 221 |
import numpy as np |
| 222 |
import matplotlib.pyplot as plt |
| 223 |
|
| 224 |
horizon = np.genfromtxt('horizon.csv', delimiter=',') |
| 225 |
horizon = horizon[1:, :] |
| 226 |
|
| 227 |
ax = plt.subplot(111, polar=True) |
| 228 |
bars = ax.plot(horizon[:, 0] / 180 * np.pi, |
| 229 |
(90 - horizon[:, 1]) / 180 * np.pi) |
| 230 |
# uncomment the 2 following lines when using -c flag |
| 231 |
# ax.set_theta_direction(-1) |
| 232 |
# ax.set_theta_zero_location('N') |
| 233 |
plt.show() |
| 234 |
</pre></div> |
| 235 |
|
| 236 |
<center> |
| 237 |
<img src="rhorizon_polar_plot.png"><br> |
| 238 |
Horizon plot in polar coordinates. |
| 239 |
</center> |
| 240 |
|
| 241 |
<h3>Raster map mode</h3> |
| 242 |
|
| 243 |
Raster map mode (output maps "horangle*" become input for <em>r.sun</em>): |
| 244 |
<div class="code"><pre> |
| 245 |
g.region raster=elevation -p |
| 246 |
|
| 247 |
# we put a bufferzone of 10% of maxdistance around the study area |
| 248 |
# compute only direction between 90 and 270 degrees |
| 249 |
r.horizon elevation=elevation step=30 start=90 end=300 \ |
| 250 |
bufferzone=200 output=horangle maxdistance=5000 |
| 251 |
</pre></div> |
| 252 |
|
| 253 |
<h2>REFERENCES</h2> |
| 254 |
|
| 255 |
<p>Hofierka J., 1997. Direct solar radiation modelling within an |
| 256 |
open GIS environment. Proceedings of JEC-GI'97 conference in Vienna, |
| 257 |
Austria, IOS Press Amsterdam, 575-584 |
| 258 |
|
| 259 |
<p>Hofierka J., Huld T., Cebecauer T., Suri M., 2007. Open Source Solar |
| 260 |
Radiation Tools for Environmental and Renewable Energy Applications, |
| 261 |
<a href="https://www.researchgate.net/profile/Ralf-Denzer-2/publication/269166795_Environmental_Software_Systems_Volume_7_Second_Edition/links/54847f450cf283750c370758/Environmental-Software-Systems-Volume-7-Second-Edition.pdf#page=464">International Symposium on |
| 262 |
Environmental Software Systems</a>, Prague, 2007 |
| 263 |
<p>Neteler M., Mitasova H., 2004. Open Source GIS: A GRASS GIS |
| 264 |
Approach, <a href="https://link.springer.com/book/10.1007/978-0-387-68574-8">Springer</a>, New York. |
| 265 |
ISBN: 1-4020-8064-6, 2nd Edition 2004 (reprinted 2005), 424 pages |
| 266 |
|
| 267 |
<p>Project <a href="https://joint-research-centre.ec.europa.eu/photovoltaic-geographical-information-system-pvgis_en">PVGIS</a>, European |
| 268 |
Commission, DG Joint Research Centre 2001-2007 |
| 269 |
<p>Suri M., Hofierka J., 2004. |
| 270 |
A New GIS-based Solar Radiation Model and Its Application for |
| 271 |
Photovoltaic Assessments. <a href="http://web.archive.org/web/20060212122544/http://www.blackwellpublishing.com:80/toc.asp?ref=1361-1682">Transactions |
| 272 |
in GIS</a>, 8(2), 175-190 (DOI: <a href="https://doi.org/10.1111/j.1467-9671.2004.00174.x">10.1111/j.1467-9671.2004.00174.x</a>) |
| 273 |
|
| 274 |
<h2>SEE ALSO</h2> |
| 275 |
|
| 276 |
<em> |
| 277 |
<a href="r.sun.html">r.sun</a>, |
| 278 |
<a href="r.sunmask.html">r.sunmask</a>, |
| 279 |
<a href="r.viewshed.html">r.viewshed</a> |
| 280 |
</em> |
| 281 |
|
| 282 |
<h2>AUTHORS</h2> |
| 283 |
|
| 284 |
Thomas Huld, Joint Research Centre of |
| 285 |
the European Commission, Ispra, Italy |
| 286 |
<br> |
| 287 |
<p>Tomas Cebecauer, Joint Research Centre |
| 288 |
of the European Commission, Ispra, Italy |
| 289 |
<br> |
| 290 |
<p>Jaroslav Hofierka, GeoModel s.r.o., |
| 291 |
Bratislava, Slovakia <br>Marcel Suri, Joint Research Centre of the |
| 292 |
European Commission, Ispra, Italy |
| 293 |
<p>© 2007, Thomas Huld, Tomas Cebecauer, Jaroslav Hofierka, Marcel Suri |
| 294 |
|
| 295 |
<a href="mailto:[email protected]">[email protected]</a> |
| 296 |
<a href="mailto:[email protected]">[email protected]</a> |
| 297 |
<a href="mailto:[email protected]">[email protected]</a> |
| 298 |
<a href="mailto:[email protected]">[email protected]</a> |