Method for Alaska (AMAK) was developed in the
NMFS Alaska Fisheries Science Center by Dr. James Ianelli using AD
Builder. This is an age based estimation model that supports
multiple fisheries and sparse data availability.
The AMAK model is an explicit age-structured model that uses a
forward projection approach and maximum likelihood estimation to solve
for model parameters. The operational population dynamics
model is defined by the standard catch equation with various
modifications such as those described by Fournier and Archibald (1982),
Hilborn and Walters (1992) and Schnute and Richards (1995).
The model simultaneously fits parameters for up to seven
components of a penalized likelihood function. The likelihood
function includes parameters for
proportion at age
fishing mortality regularity
prior distributions on
natural mortality and survey catchability.
Survey and fishery selectivity are modeled similarly using one
of 4 alternative functional forms and a time-varying curvature penalty
which allows the model parameters to change over time as well as age.
Fishery age composition, modeled as a multinomial distribution,
requires a measure of effective sample size for each year. The
posterior distribution of the estimated
parameters can be approximated using the Markov Chain Monte Carlo