<div dir="ltr"><div><div><div><div><div><div><div>r.los is based is based on the US Fire spread model (BEHAVE), or at least it was a few years ago (10? 15? Augh!). It is not based on the Canadian model (FBP).<br><br></div>
The two models are significantly different in their parameters, and they don't model the same things. For what it is worth, the r.los model models the question "How long will fire take to spread from that other burning neighbour cell to me?", so it ends up being an event-driven system, not a time-driven system.<br>
<br></div>Having said that, it's a good point that looking at that code could be informative.<br><br></div>I'm personally most interested in taking a page from the model built for the prometheus project where they did the fire processing in a vector model, which seems to reduce the nearest neighbour spread issue. They published a paper on that, but I don't have it to hand. The idea was that each point in a tiny circle around the start point would have a fire prediction made, and then the assembled ellipses would be dissolved and the new fire front would emerge as the collection of unique points on the outside of the fire. <br>
<br></div>The topography and fuels are all straight point samples from rasters. The weather is from a common hourly weather data set.<br><br></div>The advantage of the vector system is that the raster leads to strong pulls in the direction of the raster grain (usually NS and EW). If you design it for an 8-neighbour system, there are other effects. You could do a six-neighbour system with a hexagonal "raster". All of them lead to effects from the grain of the raster. These effects are exacerbated with a time-driven system.<br>
<br></div>Anyway, I'm obviously off on a discussion that has a time and place elsewhere... :-)<br><br></div>Cheers,<br>Angus.<br><div><div><div><div><br> <br></div></div></div></div></div><div class="gmail_extra"><br>
<br><div class="gmail_quote">On Thu, Nov 21, 2013 at 2:41 AM, Blumentrath, Stefan <span dir="ltr"><<a href="mailto:Stefan.Blumentrath@nina.no" target="_blank">Stefan.Blumentrath@nina.no</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div link="blue" vlink="purple" lang="NO-BOK">
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d">Hi Tom<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d" lang="EN-US">Besides the technical aspects, just in case you do not know them yet, you may probably get some scientific inspirations from e.g. r.los (and related
modules) in GRASS (see <a href="http://grass.osgeo.org/grass64/manuals/r.ros.html" target="_blank">
http://grass.osgeo.org/grass64/manuals/r.ros.html</a>) which represent existing GIS models for wildfire spread. Maybe something to build on…?<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d" lang="EN-US"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d" lang="EN-US">Cheers<span class="HOEnZb"><font color="#888888"><u></u><u></u></font></span></span></p>
<span class="HOEnZb"><font color="#888888">
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d" lang="EN-US">Stefan<u></u><u></u></span></p>
<div>
<p class="MsoNormal"><u></u> <u></u></p>
</div>
</font></span></div>
</div>
</blockquote></div><br></div>