estSigmaR {scape}

Estimate Recruitment Sigma

Description

Estimate sigma R (recruitment variability), based on the empirical standard deviation of recruitment deviates in log space.

Usage

estSigmaR(model, digits=2)

Arguments

model

fitted scape model containing element Dev.

digits

number of decimal places to use when rounding, or NULL to suppress rounding.

Value

Vector of two numbers, estimating recruitment variability based on (1) the estimated age composition in the first year, and (2) subsequent annual recruitment.

Note

This function uses the empirical standard deviation to estimate sigma R, which may be appropriate as likelihood penalty (or Bayesian prior distribution) for recruitment deviates from the stock-recruitment curve. The smaller the estimated recruitment deviates, the smaller the estimated sigma R.

estSigmaR can be used iteratively, along with estN and estSigmaI to assign likelihood weights that are indicated by the model fit to the data. Sigmas and sample sizes are then adjusted between model runs, until they converge. The iterate function facilitates this procedure.

If ss is the sum of squared recruitment deviates in log space and n is the number of estimated recruitment deviates, then the estimated sigma R is:

\sigma_R=\sqrt{\frac{ss}{n}}

The denominator is neither n-1 nor n-p, since ss is based on deviates from zero and not the mean, and the deviates do not converge to zero as the number of model parameters increases.

See Also

getN, getSigmaI, getSigmaR, estN, estSigmaI, and estSigmaR extract and estimate sample sizes and sigmas.

iterate combines all the get* and est* functions in one call.

plotN and plotB(..., what="s") show what is behind the sigma R estimation.

scape-package gives an overview of the package.

Examples

getSigmaR(x.cod)  # sigmaR used in assessment 0.5 and 1.0
estSigmaR(x.cod)  # model estimates imply 0.20 and 0.52

getSigmaR(x.ling)  # 0.6, deterministic age distribution in first year
estSigmaR(x.ling)  # model estimates imply 0.36

getSigmaR(x.sbw)
estSigmaR(x.sbw)  # large deviates in first year
plotN(x.sbw)      # enormous plus group and 1991 cohort

# x.oreo assessment had deterministic recruitment, so no deviates