Barents Sea fish modelling in Uncover

1
Barents Sea fish modelling in Uncover

• Daniel Howell
• Marine Research
• Institute of Bergen

2
(No Transcript)
3
Fish model

• Multispecies fish population model
• Cod, capelin, herring
• Age and length structured
• To be implemented in FLR
• Based on Gadget model

4
Gadget

• Simulation model
• Create a virtual population within the model
• Follow the fish through their lives
• Fishing, mortality, growth, maturation, etc.
• Process driven
• E.g. percentage becoming mature, not percentage
  mature at age

5
Gadget

• Agelength based
• Multiple species, stocks, fleets, areas
• Separation of model and data
• No data required for the simulation run
• Statistical functions used to compare model and
  data

6
Gadget

• Specify a model with
• Choice of equations for growth, reproduction,
  fishing selection...
• Parameters in those equations
• fixed or estimated
• Data
• Statistical functions measuring fit between model
  and data

7
Gadget

• Simulation model is made, without using the data
• Uses specified stocks, fleets, growth
  equations...
• Produces a virtual population through time, and
  virtual catches from that population

8
Gadget

• Compared model results against the real-world
  data
• Statistical functions assign a numerical score to
  each data set
• Combined in a weighted sum to give a single
  likelihood score

9
Gadget

• A one dimensional measure of the success of the
  model
• Can be used to optimize the model
• Repeat runs are made using different values of
  key parameters
• Attempting to find the lowest score
• the best match to the data

10
Optimization

• Optimize parameters
• Structure of model is fixed
• E.g. Select a dome shaped fishing selectivity
• Will remain dome shaped
• Exact shape will be optimized

11
Model Overview

• Very complex for a fisheries model
• Need to avoid adding any more complexity that is
  necessary
• Extra complexity/flexibility needs either
• Data to optimize to
• Externally derived parameters

12
Model Overview

• Monthly time steps
• Several Areas (Barents Sea and other subsidary
  areas for migration)
• Currently only considering interactions in the
  Barents Sea

13
Model components

• Multiple stocks
• Different species
• Different stocks of one species
• Split by maturity or sex
• Different genetic components?
• Fish can move between stocks
• Maturation
• Physically move between two distinct stocks

14
Model components

• Growth
• Mean length growth can be a combination of
• length, age, weight, condition, water
  temperature
• Other physical factors?
• Actual growth
• mean growth for each timestep is converted in to
  a distribution, with estimable parameters

15
Model components

• Migration
• Multiple areas
• Movement pre-specified or modelled
• Modelled as moving from one area to another
• Can vary over time (needs to be pre-specified or
  have enough data to estimate changes)

16
Model Components

• Reproduction
• Can simply estimated on a yearly basis to best
  fit the data
• Or based on mature population characteristics
• Can also include other factors (e.g. temperature)

17
Model Components

• Reproduction
• Closed life cycle is possible
• Can be based on SSB, or on the length, weight and
  possibly condition of adult fish
• Can be based on length distribution of fish, not
  just overall SSB
• Simple modelling of fish larvae possible
• But has to be on the same time scale as the fish
  model (monthly time step)

18
Model Components

• Predation
• fish can eat other stocks
• predation is length based (predator and prey)
• cannibalism is possible
• desired diet of a predator is spread over
  available prey, with preference factors
• Fishing
• fleets are treated as predators

19
Model Components

• Fishing
• Fleets can have their own selection function
• typically selecting on length
• Can either model (or specify) catch in tons
• Or model fishing mortality

20
Model Flexibility

• Parameters can be pre-specified or estimated
• Estimate
• once for all years
• separately for each year
• split years into blocks, and estimate for each
  block

21
Model Flexibility

• Choice of functions typically available
• Growth
• Fishing selectivity
• Etc.
• Can write new functions, or modify existing ones,
  relatively easily if required

22
Modelled population

• The program keeps track of the details of the
  virtual population, and outputs summary
  statistics
• e.g.
• Numbers, biomass, weight at length and age
• Catches in numbers and weight
• Predation by one species on another, by length

23
New or improved process models

• Improvements can come as either
• Processes modelled within the program
• Processes implemented as fixed parameters
• Need to keep complexity to a minimum

24
Task

• To look at ways in which the outputs from the
  different parts of WP1 and WP2 can be
  incorporated into the fish population model

25
Sub Models (1.4)

• Migration
• Model is large scale
• Migration is specified as percentage moving
  between areas in a given month (process driven)

26
Sub Models (1.3)

• Genetic/behavioural changes through time
• Have the possibility to have time dependant
  effects in most parameters
• e.g. Migration, recruitment, growth
• Not yet stock-size dependant

27
Sub Models (1.1, 2.4)

• Fecundity
• Have
• Number of fish by age and length
• Weight of fish in age/length cell
• Condition of fish

28
Sub Models (2.1, 2.2, 2.3)

• Larval growth/survival
• Can include simple larval growth
• Can have eggs produced in one area appearing as
  fish in another
• Can include environmental factors (e.g.
  Temperature)
• Cannot include very small time steps
• Needs data or pre-specified parameters

29
Recovery Scenarios

• Can set up quite detailed scenarios
• Time dependant
• good and bad years
• Dont have much time to actually do this
• Need to prioritize