3
Spread phenomena usually show uneven movement over space. Such unevenness
5
<BR>1) the uneven conditions from location to location, which can be called
6
SPATIAL HETEROGENEITY, and
7
<BR>2) the uneven conditions in different directions, which can be called
9
<BR>The anisotropy of spread occurs when any of the determining factors
10
have directional components. For example, wind and topography cause anisotropic
13
<P>One of the simplest spatial heterogeneous and anisotropic spread
14
is elliptical spread, in which, each local spread shape can be thought
15
as an ellipse. In a raster setting, cell centers are foci of the spread
16
ellipses, and the spread phenomenon moves fastest toward apogees and slowest
17
to perigees. The sizes and shapes of spread ellipses may vary cell by cell.
18
So the overall spread shape is commonly not an ellipse.
20
<P><I>r.spread </I>simulates elliptically anisotropic spread phenomena,
21
given three raster map layers about ROS (base ROS, maximum ROS and direction
22
of the maximum ROS) plus a raster map layer showing the starting sources.
23
These ROS layers define unique ellipses for all cell locations in the current
24
geographic region as if each cell center was a potential spread origin.
25
For some wildfire spread, these ROS layers can be generated by another
26
GRASS raster program r.ros. The actual locations reached by a spread event
27
are constrained by the actual spread origins and the elapsed spread time.
29
<P><I>r.spread </I>optionally produces raster maps to contain backlink
30
UTM coordinates for each raster cell of the spread time map. The spread
31
paths can be accurately traced based on the backlink information by another
32
GRASS raster program r.spreadpath.
34
<P>Part of the spotting function in r.spread is based on Chase (1984)
35
and Rothermel (1983). More information on <I>r.spread</I>, <I><A HREF="r.ros.html">r.ros</A></I>
36
and <I><A HREF="r.spreadpath.html">r.spreadpath</A></I> can be found in
43
<DD> Run verbosely, printing information about program progress to standard
47
<DD> Display the "live" simulation on screen. A graphics window
48
must be opened and selected before using this option.
51
<DD> For wildfires, also consider spotting.
58
<DD>Name of an existing raster map layer in the user's current
59
mapset search path containing the maximum ROS values (cm/minute).
62
<DD>Name of an existing raster map layer in the user's
63
current mapset search path containing directions of the maximum ROSes,
64
clockwise from north (degree).
67
<DD>Name of an existing raster map layer in the user's
68
current mapset search path containing the ROS values in the directions
69
perpendicular to maximum ROSes' (cm/minute). These ROSes are also the ones
70
without the effect of directional factors.
73
<DD>Name of an existing raster map layer in the
74
user's current mapset search path containing starting locations of the
75
spread phenomenon. Any positive integers in this map are recognized as
78
<DT><B>spot_dist=</B>name
79
<DD>Name of an existing raster map layer in
80
the user's current mapset search path containing the maximum potential
81
spotting distances (meters).
83
<DT><B>w_speed=</B>name
84
<DD>Name of an existing raster map layer in the
85
user's current mapset search path containing wind velocities at half of
86
the average flame height (feet/minute).
88
<DT><B>f_mois</B>=name
89
<DD>Name of an existing raster map layer in the
90
user's current mapset search path containing the 1-hour (<.25") fuel
91
moisture (percentage content multiplied by 100).
93
<DT><B>least_size=</B>odd int An odd integer ranging 3 - 15 indicating
94
the basic sampling window size within which all cells will be considered
95
to see whether they will be reached by the current spread cell. The default
96
number is 3 which means a 3x3 window.
98
<DT><B>comp_dens=</B>decimal A decimal number ranging 0.0 - 1.0 indicating
99
additional sampling cells will be considered to see whether they will be
100
reached by the current spread cell. The closer to 1.0 the decimal number
101
is, the longer the program will run and the higher the simulation accuracy
102
will be. The default number is 0.5.
104
<DT><B>init_time=</B>int A non-negative number specifying the initial
105
time for the current spread simulation (minutes). This is useful when multiple
106
phase simulation is conducted. The default time is 0.
108
<DT><B>lag=</B>int A non-negative integer specifying the simulating
109
duration time lag (minutes). The default is infinite, but the program will
110
terminate when the current geographic region/mask has been filled. It also
111
controls the computational time, the shorter the time lag, the faster the
114
<DT><B>backdrop=</B>name
115
<DD>Name of an existing raster map layer in the
116
user's current mapset search path to be used as the background on which
117
the "live" movement will be shown.
119
<DT><B>output=</B>name
120
<DD>Name of the new raster map layer to contain
121
the results of the cumulative spread time needed for a phenomenon to reach
122
each cell from the starting sources (minutes).
124
<DT><B>x_output=</B>name
125
<DD>Name of the new raster map layer to contain
126
the results of backlink information in UTM easting coordinates for each
129
<DT><B>y_output</B>=name
130
<DD>Name of the new raster map layer to contain
131
the results of backlink information in UTM northing coordinates for each
136
The user can run r.spread either interactively or non- interactively. The
137
program is run interactively if the user types <I>r.spread</I> without
138
specifying flag settings and parameter values on the command line. In this
139
case, the user will be prompted for input.
141
<P>Alternately, the user can run r.spread non-interactively, by specifying
142
the names of raster map layers and desired options on the command line,
145
<P>r.spread [-vds] max=name dir=name base=name start=name [spot_dist=name]
146
[w_speed=name] [f_mois=name] [least_size=odds int] [comp_dens=decimal]
147
[init_time=int (>=0)] [lag=int (>= 0)] [backdrop=name] output=name [x_output=name]
148
[y_output=name] The -d option can only be used after a graphics window
149
is opened and selected.
151
<P>Options spot_dist=name, w_speed=name and f_mois=name must all
152
be given if the -s option is used.
156
Assume we have inputs, the following simulates a spotting- involved wildfire
157
on the graphics window and generates three raster maps to contain spread
158
time, backlink information in UTM northing and easting coordinates:
160
<P>r.spread -ds max=my_ros.max dir=my_ros.maxdir base=my_ros.base
161
start=fire_origin spot_dist=my_ros.spotdist w_speed=wind_speed f_mois=1hour_moisture
162
backdrop=image_burned output=my_spread x_output=my_spread.x y_output=my_spread.y
165
1. r.spread is a specific implementation of the shortest path algorithm.
166
r.cost GRASS program served as the starting point for the development of
167
r.spread. One of the major differences between the two programs is that
168
r.cost only simulates ISOTROPIC spread while r.spread can simulate ELLIPTICALLY
169
ANISOTROPIC spread, including isotropic spread as a special case.
171
<P>2. Before running r.spread, the user should prepare the ROS (base,
172
max and direction) maps using appropriate models. For some wildfire spread,
173
a separate GRASS program r.ros based on Rothermel's fire equation does
174
such work. The combination of the two forms a simulation of wildfire spread.
176
<P>3. The relationship of the start map and ROS maps should be logically
177
correct, i.e. a starting source (a positive value in the start map) should
178
not be located in a spread BARRIER (zero value in the ROS maps). Otherwise
179
the program refuses to run.
181
<P>4. r.spread uses the current geographic region settings. The output
182
map layer will not go outside the boundaries set in the region, and will
183
not be influenced by starting sources outside. So any change of the current
184
region may influence the output. The recommendation is to use slightly
185
larger region than needed. Refer to g.region to set an appropriate geographic
188
<P>5. The inputs to r.spread should be in proper units.
190
<P>6. r.spread is a computationally intensive program. The user may
191
need to choose appropriate size of the geographic region and resolution.
193
<P>7. A low and medium (i.e. <= 0.5) sampling density can improve
194
accuracy for elliptical simulation significantly, without adding significantly
195
extra running time. Further increasing the sample density will not gain
196
much accuracy while requiring greatly additional running time.
200
<EM><A HREF="g.region.html">g.region</A></EM>,
201
<EM><A HREF="r.cost.html">r.cost</A></EM>,
202
<!-- <EM><A HREF="r.mask.html">r.mask</A></EM>, -->
203
<EM><A HREF="r.spreadpath.html">r.spreadpath</A></EM>,
204
<EM><A HREF="r.ros.html">r.ros</A></EM>
207
Chase, Carolyn, H., 1984, Spotting distance from wind-driven surface fires
208
Note INT-346, Ogden, Utah.
210
<P>Rothermel, R. C., 1983, How to predict the spread and intensity
211
of forest and range fires. US Forest Service, Gen. Tech. Rep. INT-143.
214
<P>Xu, Jianping, 1994, Simulating the spread of wildfires using a
215
geographic information system and remote sensing, Ph. D. Dissertation,
216
Rutgers University, New Brunswick, New Jersey.
219
Jianping Xu and Richard G. Lathrop, Jr., Center for Remote Sensing and
220
Spatial Analysis, Rutgers University.
222
<p><i>Last changed: $Date: 2006-04-13 21:25:42 +0200 (Thu, 13 Apr 2006) $</i>