The standard deviation about ncentroid is used as the "spread" (i.e., alternate duration) feature.
Different from Case I, the channels of Case II are subject to a larger delay spread than a Doppler spread (i.e., ε>1).
These low-frequency inter-area modes are also more spread (i.e., they involve a larger number of generators) than the other inter-area modes of the system.
As for the building-to-building fire spread, three mechanisms are considered as contributing factors of fire spread, i.e., (I) thermal radiation from fire-involved buildings; (II) temperature rise due to wind-blown fire plumes; and (III) firebrand spotting.
We use radius R to calculate the spread (i.e., standard deviation) of the distribution around the perceived landing point, or σ.
By comparing the four plots, it can also be seen that power increases as the data are less spread (i.e. larger values of κ).
Therefore, the inclusion of temporal correlation as an additional detection metric is not helping in the case of delay spread (i.e. 0.097 μ s).
The possible explanation for this can be that for a short channel delay spread (i.e. 0.097 μ s), the multi-tap channel adds negligible temporal correlation.
This is due to the differences in the eigenvalue spread (i.e., the quotient between the largest and the smallest eigenvalue) between the synthetic and measured channels.
The third column ("Rater") compares the participating teachers with regard to their level of severity/leniency, covering a 5-logit spread (i.e., between -2 and +3 logits).
Although the wind density approach is potentially useful for retrospective analyses, it focuses on a single mechanism of spread (i.e. vector dispersal by wind).
