---

The core idea of this project is that elevated risk of landslides is indicated by the river rising at a steep rate, for a sustained period of time which results in a significant overall rise in the river. This would indicate rapidly increasing soil saturation in the watershed, which indicates an increased risk of landslide activity.

But what rate of rise (steepness in the river gauge graph) and what total rise indicates elevated risk of landslide activity?

Choosing critical values

If the values chosen for critical factors are too low, then critical periods are identified much more often than slides actually occur. In the scientific discussion, these events are referred to as "false positives". If the values chosen are too high, then critical periods associated with slides are potentially missed, and critical periods are identifed too close to slide events to be of any use. A missed slide event is referred to as a "false negative"

.

So far the best values for the critical factors seem to be a rate of rise averaging 0.5 feet per hour or more, with a total rise of at least 2.5 feet. Higher values miss critical periods that are associated with slides, and lower values result in a higher number of critical periods that are not associated with slides.

Identifying critical points

To identify critical points, the analysis looks at each individual reading from the river gauge. The analysis then examines readings that have been taken over the previous 5 hours. For each reading, it asks if the average rate of rise is at least 0.5 feet per hour, and if the total rise is at least 2.5 feet. If the the answer to both of these questions is yes, the data point is considered critical.

Further reading

For further discussion of what the critical factors are and how these values were chosen, see the full writeup of the historical analysis.