| windAC {windAC} | R Documentation |
A package for calculating area correction values for fatality estimation at wind farms.
Description
Post-construction fatality monitoring studies at wind facilities are based on data from searches for bird and bat carcasses in plots beneath turbines. Bird and bat carcasses can fall outside of the search plot. Bird and bat carcasses from wind turbines often fall outside of the searched area. To compensate, area correction (AC) estimations are calculated to estimate the percentage of fatalities that fall within the searched area versus those that fall outside of it. This package provides two likelihood based methods and one physics based method (Hull and Muir (2010), Huso and Dalthorp (2014)) to estimate the carcass fall distribution. There are also functions for calculating the proportion of area searched within one unit annuli, log logistic distribution functions, and truncated distribution functions.
The two likelihood methods are the truncated weighted likelihood (estTWL) and the weighted distribution (estWD). Both use carcass distances from the turbine, accounting for unequal detection by distance, to estimate the distance distribution. Alternatively, a right triangle distribution can be used for the carcass density distribution with the max distance estimated (hullMuirMaxDistance from the regression from Hull and Muir (2010) as proposed by Huso and Dalthorp (2014).
The area correction value is calculated from the combination of the carcass distance density and the proportion of area searched at each distance. The function getProportionAreaSearched uses the sf package to do this from turbine points spatial data and search area polygons. The functions geometricRectanglePropSearchTable and geometricRoadPadPropSearchTable also calculate proportion of area searched but assuming perfect geometric shapes, meaning no spatial data is required.
Search areas are often irregular. proportionAreaSearched summarizes the area searched into the proportion of area searched with one unit annuli (or ring).
Two sets of distribution functions are available. Log logistic distribution functions (see dllog). These are a transformation of the logistic distribution and use the base R functions (see dlogis).
The second is truncation functions (see dtrunc), that provide truncation for R function distributions.
Example data sets:
carcassDistance: Example data set of carcass distances found during searches under wind turbines.hallingstad: List containing two data frames from Hallingstad et al. 2018. Data frames list carcass and detection probability data for a few wind farm studies.proportionAreaSearched: An example data set of the proportion of area searched around turbines at a wind farm.
References
Hallingstad EC, Rabie PA, Telander AC, Roppe JA, Nagy LR (2018) Developing an efficient protocol for monitoring eagle fatalities at wind energy facilities. PLoS ONE 13(12): e0208700. https://doi.org/10.1371/journal.pone.0208700
Huso, M. & Dalthorp,D (2014). Accounting for Unsearched Areas in Estimating Wind Turbine-Caused Fatality. The Journal of Wildlife Management. 78. 10.1002/jwmg.663.
Hull, C. L., & Muir, S. (2010). Search areas for monitoring bird and bat carcasses at wind farms using a Monte-Carlo model.Australasian Journal of Environmental Management, 17(2), 77-87.