Skip to content

Mortality Additional mortality Adult mortality

Ricardo Oliveros Ramos edited this page Jun 11, 2026 · 1 revision

Additional mortality

There are different ways to implement additional mortality for feeding fish:

  • ANNUAL if the mortality.additional.rate.spX parameter is defined.
  • BY_DT if the mortality.additional.rate.bytDt.file.spX parameter is defined.
  • BY_DT_BY_AGE if the mortality.additional.rate.byDt.byAge.file.spX parameter is defined.
  • BY_DT_BY_SIZE if the mortality.additional.rate.byDt.bySize.file.spX parameter is defined.

Annual mortality

In this case, additional mortality varies on a yearly basis. Three parameters control this yearly variability:

  • mortality.additional.rate.spX provides the mortality rate (in $year^{-1}$)
  • mortality.additional.rate.log.spX provides the value in log mode
  • mortality.additional.rate.shift.spX provides the years when the value changes

Caution

Only one of the mortality.additional.rate.spX or mortality.additional.rate.log.spX must be defined!

If the mortality.additional.rate.shift.spX is not defined, then mortality.additional.rate.spX and mortality.additional.rate.log.spX expect one value, which will be used for the whole simulation.

If the mortality.additional.rate.shift.spX is defined and contains $N$ values, then mortality.additional.rate.spX and mortality.additional.rate.log.spX expect $N+1$ values. For example, if the shift values are 2, 5 and the mortality rates are 0.01, 0.02 and 0.03, the first value will be used between year 0 and 2, the second value between years 2 and 5, and the third value will be used between years 5 and the last year of the simulation.

If the mortalities are provided in log, the mortalies will be converted by using an exponential function. For example, if the log mortality is $-2$, the effective mortality will be $\exp\left(-2\right)$.

Additional parameters can be provided:

  • mortality.additional.rate.multiplier.spX provides a multiplier to the mortality rate. If not defined, then the multiplier value is 1.

  • mortality.additional.rate.seasonality.file.spX or mortality.additional.rate.seasonality provides a CSV file or a time-series containing multipliers, for each time-step, to modulate the annual larva time-series. When none of these parameters are defined, the multiplier will be set to 0.

With all these parameters, the larva mortality rate for a given time-step is given by:

$$ L(t) = \frac{M \times S(t) \times F_{add}(y)}{N_{step/year}} $$

By dt mortality

If the mortality.additional.rate.bytDt.file.spX parameter is defined, then a time-series of larval mortality will be provided.

Warning

Using this method, no multiplier is provided.

Caution

This will be deprecated soon. It is strongly advised to use the constant mortality rate framework with the multiplier and the saeasonality file.

Caution

Using this method, rates are still expected in $year^{-1}$, in order to facilitate a change in time-stepping.

By class and by dt mortality

If the parameter mortality.additional.rate.byDt.byAge.file.spX or mortality.additional.rate.byDt.bySize.file.spX is provided, then the mortality rate is given as a function of time and either age (in $years$) or size (in $cm$).

The CSV file must look like this:

Time;0;5;10;1e50
0;0.1;0.01;0.003;0.001
1;0.1;0.01;0.003;0.001
2;0.1;0.01;0.003;0.001
3;0.1;0.01;0.003;0.001
4;0.1;0.01;0.003;0.001
5;0.1;0.01;0.003;0.001
6;0;0.1;0.01;0.003;0.001
7;0.1;0.01;0.003;0.001
8;0.1;0.01;0.003;0.001
9;0.1;0.01;0.003;0.001
10;0.1;0.01;0.003;0.001
11;0.1;0.01;0.003;0.001
12;0.1;0.01;0.003;0.001
13;0.1;0.01;0.003;0.001
14;0.1;0.01;0.003;0.001
15;0.1;0.01;0.003;0.001
16;0;0.1;0.01;0.003;0.001
17;0.1;0.01;0.003;0.001
18;0.1;0.01;0.003;0.001
19;0.1;0.01;0.003;0.001
20;0.1;0.01;0.003;0.001
21;0.1;0.01;0.003;0.001
22;0.1;0.01;0.003;0.001
23;0.1;0.01;0.003;0.001

The first column indicates the time-step, even though it is not explicitely used by Osmose. The first row provides the class values (in $years$ or $cm$ depending on whether age or size class is used).

Caution

We strongly advise that the first class is 0 and the last class is $\infty$, to insure that all the schools fall within one class or another

Clone this wiki locally