Compute a CPUE-based indicator for use in empirical harvest control rules

cpues.ind computes a relative abundance indicator from one or more CPUE or survey indices (idx), to be used by empirical indicator (ind) modules in an mse Management Strategy Evaluation loop. Indices are first restricted to a recent window of nyears, rescaled by dividing by their mean over a set of reference years (refyrs), and then combined across indices, using weights, with one of three methods: a weighted rescaled mean ("mean"), a weighted mean of standardized z-scores on the log scale ("zscore"), or a weighted mean of smoothed, rescaled indices ("smooth"). For combine = "mean", the function additionally computes the mean level and log-linear slope of the combined indicator over the recent window.

Usage

cpues.ind(
  stk,
  idx,
  refyrs,
  nyears = 4,
  indices = names(idx),
  combine = c("mean", "zscore", "smooth"),
  weights = rep(1, length(indices)),
  enp.mult = 0.2,
  args,
  tracking
)

Arguments

stk: An FLStock object, returned unmodified in the output list.
idx: An FLIndices object with the CPUE or survey indices from which the indicator is computed.
refyrs: Reference years used to rescale ("mean", "smooth") or standardize ("zscore") each index.
nyears: Number of recent years, ending in dy, over which the indicator is computed. numeric, defaults to 4.
indices: Names of the elements of idx to use. Defaults to names(idx), i.e. all indices in idx.
combine: Method used to combine the rescaled indices into a single indicator, one of "mean", "zscore" or "smooth".
weights: Weights applied to each index, used by all three combine methods, passed on to weighted.mean. Defaults to equal weights, numeric of length length(indices).
enp.mult: Smoothing parameter passed to smooth_index as enp.mult, only used when combine = "smooth". Defaults to 0.2.
args: A list of MSE loop arguments, e.g. ay (assessment year) and dy (last data year), unpacked using mse::spread.
tracking: An mse tracking object (FLQuant) used to store the computed indicator(s) for later inspection.

Value

A list with three elements:

stk: The input FLStock, unchanged.
ind: An FLQuants with the combined indicator(s): ind, mean and slope for combine = "mean"; mean only for combine = "zscore" or combine = "smooth".
tracking: The updated tracking object.

Details

cpues.ind is designed to be called inside an mse indicator (ind) module, and expects args to contain at least the current assessment year (ay) and last data year (dy), unpacked internally via mse::spread. The mean (and, for combine = "mean", the slope) of the indicator is stored in tracking under "mean.ind" and "slope.ind".

Combination across indices is done by the weighted.mean method for FLQuants, which by default excludes NA values from both the weighted sum and the sum of weights, so indices missing in a given year do not bias the result towards zero or dilute the contribution of the indices that do have data.

Author

Iago Mosqueira (WMR), FLR Team.

Examples

if (FALSE) { # \dontrun{
data(ple4)
data(ple4.indices)

args <- list(ay=2017, dy=2017)
tracking <- FLQuant()

out <- cpues.ind(stk=ple4, idx=ple4.indices, refyrs=2005:2010,
  nyears=4, combine="mean", weights=c(1, 2), args=args,
  tracking=tracking)
} # }