fatalityCMR Q and A
------------------------------------------------------------------------- Probability of not finding a carcass given it is in surveyed area (P0): ------------------------------------------------------------------------- Uniform entry rate Pulsed entry rate SE(unif) SE(pulse) Type1 0.3178 0.3174 0.0991 0.0991 Type2 0.3178 0.3174 0.0991 0.0991 -------------------------------------------------------------------------OK, this one I think I understand, with P0 having same meaning as in our paper, right?
==Yes.
Maximum number of fatalities (with risk threshold 0.05 ) ------------------------------------------------------------------------- Uniform entry rate Pulsed entry rate SE(unif) SE(pulse) Type1 2.6135 2.6103 0.8234 0.8219 Type2 2.6135 2.6103 0.8234 0.8219 -------------------------------------------------------------------------labeling causes me to interpret this as follows:
Is my interpretation correct?
Also, is this conditional on the observed number of carcasses?
So if you really observed 1 carcass, would I rewrite above as:
Pr(number of fatalities > 2.6135 | 1 carcass observed) <0.05
== Here, knowing that x carcasses have been observed, I compute Nmax that verifies
Pr(number of fatalities = Nmax | x ) = 0.05
So in the example,
Pr(number of fatalities = 2.6135 | 1 carcass observed) = 0.05
Ad-hoc estimate of the number of fatalities ------------------------------------------------------------------------- Uniform entry rate Pulsed entry rate SE(unif) SE(pulse) Type1 0.6504 0.6487 0.5938 0.5918 Type2 0.6504 0.6487 0.5938 0.5918 -------------------------------------------------------------------------I don't know what the above labeling means: "ad-hoc estimate of the number of fatalities".
==Sorry about the text. Throughout, if any of you can think of better labels and text,
feel free to change.
This "ad hoc" estimate is the sum over n of [n * Pr(number of fatalities = n | x carcasses observed)]
I stopped the sum at a finite maximum n, so divided the whole thing by sum over n of [Pr(number of fatalities = n | x carcasses observed)]
I realize I should have run this by you, let me know if clarifications are needed
Extrapolations to whole wind farm --------------------------------------- ------------------------------------------------------------------------- Maximum number of fatalities (with risk threshold 0.05 ) ------------------------------------------------------------------------- Uniform entry rate Pulsed entry rate SE(unif) SE(pulse) Type1 2.6135 2.6103 0.8234 0.8219 Type2 2.6135 2.6103 0.8234 0.8219I assume this is based on simple area expansion. Is variance computed by treating area searched as a known constant, I assume? So if you searched 50% of turbines, then var for entire area estimate would be 4*var(number fatalities in searched area)?
==Yes,
this is a simple multiplication by a correction factor that is derived
from the proportion of turbine searched and the proportion of the death
zone around each turbine that is searched.