Latest
SS News: March 26, 2020
Greetings
Stock Synthesis (SS) Users,
This
release of SS (3.30.15.00) encompasses nine months and nearly 50
commits of augmentations and fixes. Of particular note are new
capabilities to read a random seed for generating bootstrap data,
allow users to skip forecasting, improve Dirichlet-multinomial
compositional data weighting output, and improve the usability of
the semi-parametric (2D-AR) selectivity option. Most fixes are for
seldom used SS features including changes so that Lorenzen natural
mortality can now be used when time-varying growth is estimated, and
corrected calculations for the expected values of morph
composition. It always is advisable to update your version to take
advantage of improvements in outputs and to learn of new options.
The full list of changes is in the change
log and searchable by keyword, date, and version. An
abbreviated, recent list is posted at the end of this message.
Thank you to all users who reported bugs or requested features.
The latest
3.30.15.00 executables can be found in the SS
document library. Please note that between releases, fixes
are posted in the latest
executables folder. Sometimes the Linux and OSX version are
not compiled as frequently as the Windows version, so please do not
hesitate to contact the SS development team (“SS team”) at nmfs.stock.synthesis@noaa.gov
if you require a more up-to-date Linux or OSX version.
An
essay describing how fishing mortality, F, is used and reported by
SS is attached to these release notes. This essay was written to
answer questions users had about the various aspects of F in SS. It
is also included in the SS manual.
Website
changes: The SS team has been making SS more accessible. As part of
this process, a
copy of the SS source code (the ADMB tpl) for this release is
available to browse. In addition, the SS
forums are now public, meaning that even those without an
account can view and create posts. Need a refresher on creating
forum posts or (for registered users) subscribing/unsubscribing to
the forums? See this article on using
forums in the vlab user guide.
Update
on GUI - Stock Synthesis Interface (SSI): New features added to the
SSI
expand its capabilities and make the interface more user-friendly.
These features include the ability to choose a default model to load
when starting SSI, selectivity and stock recruitment relationship
curves can be viewed while editing the model, reading models with
tagging data, and r4ss plots can be created with a click of a button
after a successful model run. There is also a new quick
start guide for the SSI which documents initial setup.
ss3sim
updates: ss3sim (on CRAN
and github) has been
updated to work with SS 3.30 models. ss3sim is an R package that
facilitates flexible, rapid, and reproducible fisheries stock
assessment simulation testing with SS. If you are interested in
using ss3sim and wish to develop additional functionality, please open an issue in
the ss3sim repository. The ss3sim team is happy to coordinate
with you to make the range of options that ss3sim can handle more broad.
r4ss
updates: The r4ss functions for reading and writing SS input files
(see functions SS_readctl(), SS_writectl(), SS_readdat(), and
SS_writedat() especially) have been expanded to work with a wider
range of SS models. If you are using R as a tool to interact with
SS, these functions may be of use for modifying SS files
programmatically within R. The SS team is using these functions
within ss3sim, to advance a decision support tool that provides a
GUI for control of some forecast features, and to develop a generic
MSE framework for SS. More information on changes to r4ss is
available by looking at recently
modified issues on the r4ss GitHub issues page.
Additional
tools for SS updates: An R package to use SS models in an MSE is
under active development.
SS
User Manual updates: The manual has been reorganized with three new
sections that focus on advanced SS configurations, detailed
information on internal SS processes, and general population dynamic
modeling advice. Additionally, the manual now includes a new
description of the SS bootstrapping feature and an improved
description of how to use conditional-age-at-length in SS. Have a
suggestion for the manual or need clarification? Email the SS
development team at nmfs.stock.synthesis@noaa.gov.
Contact
us: Please do not hesitate to report strange SS behavior, ask a
question about SS, or request a feature by posting on the forums
or emailing the SS team at nmfs.stock.synthesis@noaa.gov.
Please contact VLAB administrators at vlab.admin@noaa.gov to change
your username or password.
Change
log since last release:
-
Dirichlet-multinomial
composition likelihood: Improve reporting of adjusted sample
sizes; include adjusted sample sizes in bootstrap data
generation. NOTE: still need to extend to generalized size comp.
-
empirical
wt-at-age: Fix internal storage of annual fecundity; improve
filling wtatage in forecast years; merge writing of
wtatage.ss_new into other reporting rather than separate iteration
-
Sdreport:
improve input format for extra sdreport quantities and introduce
sdreport output for M-at-age
-
Selectivity:
introduce ability to set selex = 0.0 for young age range because
length selex alone can allow their selection
-
Bootstrap
data: allow user input (starter.ss) of seed for random number generator
-
Morph
composition data: fix major problem that affected expected
values for morph composition since creation of this feature in
3.24; improve output
-
Forecast:
add no forecast (-1) option and improve simple one year forecast
option (0)
-
Priors:
gamma prior was incorrect, had wrong sign
-
Priors:
clarify how prior like for non-estimated parameters is used,
esp. for prior on survey Q
-
Data
Input: fix problem that converted fractional data months to integers
-
Forecast:
make forecast F robust to high catches and pin at maxF; improve
performance when used fixed F inputs
-
Report:
output each block's age-age' key when blocks used for estimated
age-age' key
-
Selectivity:
clarify age/len terminology in parameter names and warnings for 2DAR
-
Warnings:
improve accuracy of N warn count; add warnings for illogical
data specifications for retention; improve warnings for
%depletion fleet
-
Retention:
add handling of age-retention when discard option=3; add check
for correct parameter range for asymptotic selectivity
-
Internal
calc: revise the nil expected value warning so it only adds a
smaller constant
-
Bootstrap:
can generate observations in forecast out to 50 years
-
Warnings:
note that F0.1 search can fail if bycatch fleets' catch is included
-
Warnings:
make code more robust to retained catch data used when fleet is
bycatch only
-
Growth:
change internal order of operations and time index tracking to
allow Lorenzen M to work with time-varying growth
-
Report:
add N observation info to logL table in report.sso
-
Forecast:
option to set max bias adjustment to 1.0 for all years was
incorrectly beingapplied in forecast
-
Control
input: revise autogen to retain input values in control.ss_new
unless used; also fix autogen of block offsets
-
Report:
formatting fixes in .ssnew and reporting in Impl_err and F
parameters in output
-
Report:
fix output issue with recruitment in time_series for
multi-season models
-
Report:
fix Z-at-age output for N morphs>1
-
Input:
fix the mirroring of tag recapture parameters
-
Forecast:
correct problem with heterogeneous catch types (bio vs num) in forecast
-
Report:
fix output of sample size in data.ss_new for mean length@age observations