December 2, 2009
Well, we almost got through Thanksgiving without a landslide. There was another landslide on Chuckanut Drive near Spokane Street. The first time I saw reported was 11:45am on 11/26/2009. That makes 11 reported landslides so far.
These little storms that roll through are really a wealth of information. Since we are embarking on trying to make a landslide forecasting system, we need to figure out the precipitation thresholds. The nice thing is, these small landslides are probably failing just at the border of that threshold. The problem is figuring out the precipitation variables. As far as I can figure, it doesn’t look like the USGS Seattle Landslide threshold model is holding up in the rest of the state (nor should we expect it to).
So, I threw together the precipitation data for a 3-day, 15-day USGS threshold models. We are assuming that the areas are probably at or very close to antecedent moisture threshold values. As you can see in the map, almost all of western Washington was above the threshold and guess what, we did have landslides. However, the amount of false positive that this map portrays is fairly concerning, unless there are a ton of landslides we haven’t seen. I guess the model is fairly simple in that it just says there is a likely chance that landslides could occur past the threshold. I think we might be able to improve upon this, maybe we can come up with a threshold model and as we pass the threshold, we can come up with a estimates landslides per acre, or something relevant that we could better plan or manage for an impending disaster. The biggest hurdle I am facing right now is, what sort of day ratio is the correct one. I don’t think the 3 day-15 day is going to work, or if it is, we need to drastically change the threshold value for various parts of Washington. Second, do we have indicator areas when the thresholds are starting to get reached, say, Chuckanut Drive or Eagle Cliffs down on SR 4. I have a dozen other questions, both those are foremost on my mind today.