SS usage questions

RE: likelihood values

2024-12-13T14:12:32Z

Hi Marta,

I would actually suggest that you post this question to the ss3 google group as most people have moved to that platform and are generally quicker to answer questions there.

likelihood values

2024-12-13T08:52:00Z

Dear SS Experts,

First of all, sorry for asking too many questions, but we are working on improving our SS model for hake, and in this process many doubts and questions arise.

We are currently experimenting with the lambda parameter to control the contribution of each data source to the total likelihood. During this process, we have been reviewing the table of likelihood values (see image) and have noticed some points that we don't fully understand:

For example, in the case of catches, the likelihood value in the table correspond to the sum of the "Like" column from the catch information table (which we understand contains the -log L for each observation). This makes sense to us.

However, for other components such as surveys or discards, we observe something different. Specifically, the likelihood value for the indices in the likelihood table is -79.5787, and does not correspond to the sum of the 'Like' column in the indices table. Instead, it corresponds to the sum of the "Like" column plus log(SE). The resulting value changes significantly from 112.1084 to -79.5787 when log(SE) is added.

We do not understand why the "Like" column, which we interpret as -log(L), is modified by adding log(SE) for components such as discards or surveys. This additional term seems to change the contribution of these components to the total likelihood significantly.

We would be very grateful if you could explain the rationale for adding this term.

Thank you very much for your time and help!

RE: Seed option in started,ss

2024-11-13T08:34:42Z

Hi Rick,

Thanks for looking into this. If it would be helpful, I can also share the results of my tests - just let me know and I'll be happy to provide whatever is needed.

Kind regards,
Marta

RE: Seed option in started,ss

2024-11-12T20:08:10Z

Hi Marta,

I looked at the code and did not see an obvious reason for there to be an interaction like you are seeing. So, I will need to set up a local test case to investigate further.

If I find something that needs to be changed, there will be a github issue created here: https://github.com/nmfs-ost/ss3-source-code

Also, most of our public discussions have moved to this google forum which many users find more convenient: https://groups.google.com/g/ss3-forum

Rick

Seed option in started,ss

2024-11-12T13:54:31Z

Dear SS Experts,

I am working with a hake sex-separated SS model. During a jitter analysis I observed the following behaviour. When setting the seed in the starter file and using the "Initial Parameter Values=0" option, the seed seems to work correctly as I ran three models with the same seed and all three gave identical estimates.However, when I switch to "Initial Parameter Values=1" and set the seed in the starter file, the seed does not seem to work consistently. I ran three models with the same seed, but each gave different results. Is there any explanation for this behaviour?

Thank you for your help.

RE: Recruitment Deviation Setup

2024-04-22T12:50:41Z

Thank you both Rick and Adam for your comments and detailed explanations. They are very helpful. On the basis of these ideas we will work on the recdevs of our model.

Keep in touch,
Marta

RE: Recruitment Deviation Setup

2024-04-18T21:21:39Z

Following up on Rick's response.

I am currently undertaking two assessments and have been exploring a number of these issues. The species (NZ snapper) is relative long lived (M ~ 0.07). Both stocks have shown a marked increase in productivity over the last 10 years. The stocks have a good time series of age composition data from the mid 1970s onwards. There was a substantial amount of catch from the preceding decades although the magnitude of the catch is not well known.

Previously, we conducted a full catch history model starting from unexploited conditions. However, the change in productivity means that the B0 is no longer considered appropriate for management purposes. Starting the model under exploited conditions (in 1975) means that we no longer need to include the less reliable large catches from the earlier years.

Previously, we have tended to estimate recruitments with the constrained dev vector (option 1). Some simulation work indicated that this approach had the potential to bias estimates of stock status. I also encountered issues with the stock crashing when running MCMC - when the stock was at very low biomass the constrained rec devs were unable to generate enough recruits (for a relatively small proportion of the MCMC draws, but frustrating). This was resolved when switching to unconstrained devs (option 2). Other comparisons did not result in appreciable differences in the estimated recruitment/biomass from the two options.

Starting the model in 1975, I am estimating an initial equilibrium fishing mortality - informed by the early age composition data. However, those data exhibit considerable variability in recruitment over the 20-30 y age range. So, I have also been estimating early rec devs to better fit those age comps and incorporate the uncertainty in the initial F. If those data aren't very informative then you can get confounding between the initial F and the early rec devs. The attached plot of the rec devs for the two periods (early and main) is attached. The very early (1960-65) recruitments are highly uncertain, while the 1970-74 deviates are quite well informed by the age comp data - there is good progression of some very strong year classes. Estimating the early rec devs allowed a good fit to those data, but didn't greatly change the estimate of initial F or initial biomass.

I hope these observations are useful - this could serve as a potential case study.

Adam Langley

RE: Recruitment Deviation Setup

2024-04-17T23:30:25Z

Hi Marta. Here are some thoughts. First off, the long development of SS3 emphasized flexibility of options to meet situations as they arose. The corollary of that is that there are many features for which definitive advice on good practices is not available. Early recruitment devs is one of them.

The situation in which use of early recruitment devs is most useful is when:

long time series of catch and sorter time series of data that are informative about recdevs
desire to include uncertainty about early recdevs even if they cannot be estimated individually
desire to use the recdev option 1 for later recdevs so their mean value will closely match conditions of the estimated MSY

That lead to creation of the early recdevs which are a simple dev vector without a sum to zero constraint. An advantage of use of early recdevs is that they can be turned on in a later phase to capture their variance without slightly extending runtime associated with estimating them in early phase.

Over time, we have learned, but not definitely demonstrated, that the option 1 dev-vector has disadvantages, particularly when doing MCMC, and that the sum-to-zero constraint can produce illogical patterns in the recdevs.

The current status is that I generally recommend using option 2 for all rec_devs and not using early recdevs.

Love to see someone do a good comparison of options 1 and 2 (and 3 and 4).

Rick

RE: Recruitment Deviation Setup

2024-04-17T10:50:09Z

Thanks for your reply Ian. As you point out, the Recdev page of the model output does indeed provide suggestions for the values of these parameters. However, we would like to know more about the rationale behind this in order to solve questions that we have such as the specific ones described in our first post, for example the role of the Early Recruitment Deviation Start Year.

Thank you for your help.

RE: Recruitment Deviation Setup

2024-04-11T12:09:47Z

Hi Marta Cousido,

The most common approach to set these input values is to run the r4ss::SS_plots() function in R (described at https://r4ss.github.io/r4ss/articles/r4ss-intro-vignette.html) which calls on the SS_fitbiasramp() function and provides some estimated values to set these 5 inputs. An example of that output is at the bottom of this page: https://pfmc-assessments.github.io/petrale/r4ss_plots_pre-review/_SS_output_RecDev.html. The "estimated alternative" inputs provided by that function are not always better than what a person can improve the fit over the simple algorithm (as was the case in that example), but are a useful starting place.

Let me know if you have more questions about this.

-Ian

Recruitment Deviation Setup

2024-04-11T09:16:47Z

Dear SS experts,

I am working with a sex-specific length-based model for hake and I have a question regarding the Recruitment Deviation Setup. To give some context, the model data starts in 1960 and the length-frequency information starts in 1982, before which only catch data are available. The current setup of recruitment deviations is shown below (the last four values were established following the SS suggestion in the html RecDev page).

1983 # first year of main recr_devs; early devs can preceed this era
2022 # last year of main recr_devs; forecast devs start in following year
-10 #_recdev_early_start (0=none; neg value makes relative to recdev_start)
1965 #_last_yr_nobias_adj_in_MPD; begin of ramp
1989 #_first_yr_fullbias_adj_in_MPD; begin of plateau
2019 #_last_yr_fullbias_adj_in_MPD
2023 #_end_yr_for_ramp_in_MPD

As see above there are several parameters that need to be defined in the Recruitment Deviation Setup. The first two are the main recruitment deviations start and end year. I understand that the main recruitment deviations should start when there is meaningful data to support it, for example in our case it can start when the length frequency distribution starts and we can end the period in the last year of the model. However, we also have more parameters:

Last year with no bias adjustment.
First year with full bias adjustment
Last year with full bias adjustment
First recent year with no bias adjustment

I understand that the years with full bias adjustment are supposed to be the years where there is enough information to do this, but what is enough information to do this? I see that the default is start year Nages because of the cohorts information, is there other hints to establish it? Another doubt is why the last year without bias adjustment can be before the main recruitment deviations start year?
Other doubt is the role of the Early Recruitment Deviation Start Year because I read Method (2011) and this clarify how Main Recruitment Deviations are geneared but I don't manage to understand the role of the Early?

Thank you very much for helping, if you can provide any ideas regarding this, we really appreciate that.

Methot, R.D., Taylor, I.G. and Chen, Y. 2011. Adjusting for bias due to variability in estimated recruitment in fishery assessment models. Canadian Journal of Fisheries and Aquatic Sciences 68(10): 1744–1760. doi:10.1139/f2011-092.

RE: error reading control file

2024-03-22T08:18:49Z

Thank you very much for your quick reply.

I tried the SS_read() function after looking at the echoinput.ss file. That's how i knew that it was something about the block parameters.

Unfortunately i couldn't make up by myself what the issue was.

I haven't thought much about the forecast so for the time being i did it like this.

Thanks,

Laura

RE: error reading control file

2024-03-21T18:43:36Z

Hi Laura,

Thanks for posting your question here so that other users can learn from the situation.

You were trying to debug using the SS_read() function, which does not have as complete a set of debugging tools as SS3 itself. Whenever I get a input file read error I always start by opening echoinput.sso in my text editor and scrolling down input the echoed input starts to not make sense. Actually, I usually scroll up from the bottom.

In this case I see it echoing back the selectivity parameters correctly then it shows a value of "-4" for the 2DAR switch. That is wrong and everything after that is garbled.

The problem is that you included block parameters starting with the line with the "-4", but earlier in the selectivity parameters you did not include a request for use of blocks with any of those parameters. So, you just need to add the block value (1) and your desired block function (aka block type) for the three selectivity parameters that will use blocks.

I also noticed that the block year ranges ended in 2022. This means that the model will revert to the base parameter value for the forecast. You may want. If not, put 9999 for the end year and SS3 will replace it with the end year of the forecast.

Rick & SS3 Team

error reading control file

2024-03-21T16:17:21Z

Hello, I'm trying to setup a model with selectivity blocks (time varying) but something is wrong and I can't really figure it out.

When I use the SS_read() function it gives me some indications but I believe that the issue is with reading the file itself.

It seems that it is reading the wrong inputs.

I attach the files so that you can test it.

I would appreciate your help.

Thanks

....

age_selex_parms,i=1091
Use_2D_AR1_selectivity,i=1092 ;-4
TG_custom,i=1093 ;4
Error in 1:ctllist[["N_tag_groups"]] : argument of length 0

RE: linear growth period

2024-03-08T13:39:17Z

Thank you both for your answers, we now understand growth more clearly.

RE: linear growth period

2024-03-05T17:44:41Z

Hi Marta. For Amin, it is applied to each settlement separately, so for the month 7 settlement, they will be age 0.5 the following Jan 1.

Richard D. Methot Jr. Ph.D.
NOAA Fisheries Senior Scientist for Stock Assessments

Mobile: 301-787-0241 (primary)

G. Voice: 425-666-9893‬

On Tue, Mar 5, 2024 at 9:23 AM Ian Taylor <VLab.Notifications@noaa.gov> wrote:

Hi Marta Cousido Rocha,

This is the fault of the r4ss plot created by the SSplotBiology() filtering to only show values associated with season 1, even in models with multiple seasons. I'm not sure why I chose to set it up that way long ago instead of showing the growth across seasons. The code is here: https://github.com/r4ss/r4ss/blob/main/R/SSplotBiology.R#L183.

If you look at the table in Report.sso with header "Biology_at_age_in_endyr" (returned by r4ss::SS_output() as $endgrowth) and sort by season within each year, you should be able to see the seasonal growth patterns.

I will add refinement of the plot to show seasons to the r4ss wishlist.

-Ian

--
Ian Taylor Stock Synthesis Virtual Lab Forum https://vlab.noaa.gov/web/stock-synthesis/public-forums/-/message_boards/view_message/35690554 VLab.Notifications@noaa.gov

RE: linear growth period

2024-03-05T17:23:37Z

Hi Marta Cousido Rocha,

This is the fault of the r4ss plot created by the SSplotBiology() filtering to only show values associated with season 1, even in models with multiple seasons. I'm not sure why I chose to set it up that way long ago instead of showing the growth across seasons. The code is here: https://github.com/r4ss/r4ss/blob/main/R/SSplotBiology.R#L183.

If you look at the table in Report.sso with header "Biology_at_age_in_endyr" (returned by r4ss::SS_output() as $endgrowth) and sort by season within each year, you should be able to see the seasonal growth patterns.

I will add refinement of the plot to show seasons to the r4ss wishlist.

-Ian

linear growth period

2024-03-05T08:19:32Z

Dear SS experts,

I am working with a sex-specific length-based model for hake and have a question about growth.

More specifically, the question is about the parameters that control the linear growth period.

In the control file, I set "_Age(post-settlement)_for_L1 (Amin)" to 0.5 and then let L1 be estimated. The model estimates L1 to be 20.21. However, when we look at the plot (attached), the growth is linear up to an age beyond 0.5, and the associated length is not the value of 20.21 that was estimated for L1.

I've read that Amin should be based on real age and not calendar age, but I believe it should be correct as it is, since in our case the spawning month is 1, and we have two settlements in months 1 and 7. I understand that Amin should be set relative to the first, right? Then I understand that Amin at 0.5 means that the fish will have length L1 in July and the growth will be linear until then.

Many thanks in advance,

Marta Cousido Rocha

RE: Model crash after successful read of input

2024-01-26T20:57:41Z

Thanks so much, Chantel! This is a big help. I really appreciate you taking the time to look at my files.

Best,

Laura

From: Chantel Wetzel <VLab.Notifications@noaa.gov>
Sent: Friday, January 26, 2024 3:54 PM
To: VLab.Notifications@noaa.gov
Subject: [EXTERNAL] [Stock Synthesis - SS usage questions] Model crash after successful read of input

This email has been received from outside of DOI - Use caution before clicking on links, opening attachments, or responding.

Hi Laura,

The model fails due to the input years inside the recruitment deviation set-up in the control file. The specified first year and last year of main recruitment estimation needs to fall within the start and years of the model specified at the top of the data file. Currently, both of the start and end years are before and after the model years. Additionally, the bias adjustment years should also fall within the model years. Once these years are changed, the model runs but runs into other input issues with information output in the warning file. I will let you work through those on your own.

I determined that the recruitment deviation section was the problem by opening the echoinput file which shows the last thing the model successfully checks prior to crashing was the SR section. This indicated to me that the issue likely was in the next section in the control file being checked with is the recruitment set-up section. Often when you get one of these errors where the files are read correctly but the model crashes is because a value being specified does not make sense in the context of the model set-up (years of recruitment estimation falling outside of the specified model years).

Hopefully, this is helpful.

Chantel Wetzel

--
Chantel Wetzel Stock Synthesis Virtual Lab Forum https://vlab.noaa.gov/web/stock-synthesis/public-forums/-/message_boards/view_message/34895061 VLab.Notifications@noaa.gov

RE: Model crash after successful read of input

2024-01-26T20:54:23Z

Hi Laura,

Hopefully, this is helpful.

Chantel Wetzel

Model crash after successful read of input

2024-01-26T18:30:53Z

Hi all,

I am using SS v3.30.22. The output in the command prompt suggests all the input files have been read correctly but then it crashes and gives the following error:

lower index greater than upper index in dvar_vector:: dvar-vector(const predvar_vector&)

I would appreciate any assistance in diagnosing the problem. All the input files are attached.

Best,

Laura

RE: Question on how fish age in SS3

2023-12-12T00:40:28Z

Thanks Rick, The model is set up with an annual time step and four seasons, and age specific natural mortality. Fish recruit in Q3 and then go through fishing and natural mortality for Q3 and Q4. In the numbers at age file there are 0 age-zero fish in Q1 and Q2. So I guess that might answer my own question since there are no age zero fish in Q1 and Q2, so they would all have turned one on Jan-1 and are undergoing age-1 natural mortality. Thank you!

Michelle

RE: Question on how fish age in SS3

2023-12-06T18:35:39Z

Hi Michelle,

Thanks for using the Forum to post this question.

It probably is the former, but I need a little more info before responding. Is your model annual time step with recruitment in July, or do you have explicit seasons? Are you using natural mortality option 3=agespecific; or 4=agespec_withseasinterpolate? Or perhaps you are using Lorenzen M.

Suggest you look in report.sso for this table: Natural_Mortality report:43

Rick

Question on how fish age in SS3

2023-12-06T01:49:57Z

It is my understanding that all fish age to the next year class on Jan-1 in an SS3 model. If, for example, fish are born in July, they are subject to half a year of age-0 natural mortality. Do they then experience age-1 natural mortality starting Jan-1 of the next year or do they experience age-0 natural mortality until they turn 1 in July of that year?

Thanks!
Michelle Sculley

RE: Constant Catch Projections

2023-03-22T23:51:39Z

You should be able to test this by using small constant catch values.

I'm reminded that for previous assessments we had to compare the time series of catch by fleet to the input catch values to flag output in our decision tables where the full input catch couldn't be removed (e.g. table viii in https://www.pcouncil.org/documents/2021/12/status-of-lingcod-ophiodon-elongatus-along-the-northern-u-s-west-coast-in-2021-december-2021.pdf/).

This function helps to extract the forecast catch by fleet: https://github.com/r4ss/r4ss/blob/main/R/SS_decision_table_stuff.R.

RE: Constant Catch Projections

2023-03-22T23:37:09Z

Thanks Rick,

I think I understand. There just aren't enough fish left in the stock to catch at the Btgt levels, which is why the catch doesn't match. Thanks!

Michelle

RE: Constant Catch Projections

2023-03-22T23:00:20Z

Hi Michelle,

I think the issue is that in the first year of the forecast the stock biomass is already below the biomass at Btarget.

Ratio_SSB/B0_as_target
Ratio_target	0.152
Ratio_calc	0.152
SPR@Btgt	0.183678
Fmult	3.69795
ann_F	0.588667
Exploit(Catch_dead/B_smry)	0.367554
Recruits	426.62
SPBio	3596.65

But SSB in 2021 is less at 3449. The F calculation does not use SSB directly, but presumably the age-specific abundance is not enough to get the fixed input catch out of the abundance without going to a higher F level which drives abundance down further and even less able to match the input catch.

RE: Constant Catch Projections

2023-03-22T22:17:42Z

Hi Michelle,

I have done constant catch forecasts in SS3 by simply providing fixed values at the bottom of the forecast file as you've done, but not setting anything related to caps or allocations (lines 42-62 in your file).

Hopefully Rick can figure out why the model isn't picking up on those fixed catches.

-Ian

RE: Constant Catch Projections

2023-03-22T22:02:12Z

Hi Rick,

Thanks for your quick response. Yes, If I set the forecast catch to be 5446 mt which is the estimated Dead_Catch_Btgt, then the forecast catch tends to be much lower. I've attached the forecast_report file and I'll email you the full model input files as well. If I remove the catch caps, then SS ignores my input allocations entirely.

Michelle

RE: Constant Catch Projections

2023-03-22T21:40:30Z

Also, my first guess is that having so many catch caps is interfering with keeping estimated catch at the input catch level. Have you treid removing the catch caps and relying only on the input catch levels?

RE: Constant Catch Projections

2023-03-22T21:29:32Z

Hi Michelle. We should be able to get this working and not clear why SS3 is not matching the input forecast catch values. Can you send your complete set of input files to nmfs.stock.synthesis@noaa.gov. And/or post your forecast-report.sso file here?

Also, I do not completely understand your question about setting catch to the equilibrium catch associated with Btgt. Do you mean for the forecast also?

Rick

Constant Catch Projections

2023-03-22T21:13:32Z

I'm trying to produce constant catch projection scenarios using the SS forecast system. I have attached my forecast file, but in summary I have included a maximum catch series for each fleet, and input the quarterly catch by fleet. The forecast is set to the biomass target and there is no control rule. In the example attached the maximum annual catch should total 450 mt, and the total annual catch by fleet/quarter should be 438mt. However, the SS output is producing annual catch during the forecast years of ~462mt.

Additionally, I would like to set catch to the biomass target or near the biomass target (5438 mt is Btgt Catch), but the output only produces catches around 3500mt per year.

Can someone help explain:
1. Is it possible to produce constant catch scenarios in SS forecast as opposed to constant F?

2. Why is Forecast catch (i.e. Forecatch in the report file) above the maximum catch input?

3. Do I need both the catch by fleet/year/quarter and the max annual catch inputs?

4. Am I missing any other settings that might be causing problems?

Thank you!
Michelle Sculley

RE: Trouble with multiple growth morphs

2022-11-09T22:46:26Z

Rick,

So the biomass is aggregated across morphs and then recruitment gets calculated using the SRR and then is apportioned based on the apportionment parameters. Thanks for the clarification.

Cheers,
Thom

RE: Trouble with multiple growth morphs

2022-11-09T22:36:07Z

The morph composition feature was first developed for a 3 area model with one morph spawning in area 1 and another in area 3, then with seasonal mixing in area 2. Otolith microchemistry was used to get proportions in area 2. A shortcoming of SS3 is that it only allows a single global spawner biomass to be used in the spawner-recruitment relationship.

On Wed, Nov 9, 2022 at 2:29 PM Thom D. Teears <VLab.Notifications@noaa.gov> wrote:

Ian,

Thanks for the information. I appreciate it. That seems like a valuable feature if the data exists to incorporate into the assessment.

Cheers,

Thom

--
Thom D. Teears Stock Synthesis Virtual Lab Forum https://vlab.noaa.gov/web/stock-synthesis/public-forums/-/message_boards/view_message/25908001 VLab.Notifications@noaa.gov

RE: Trouble with multiple growth morphs

2022-11-09T22:29:48Z

Ian,

Thanks for the information. I appreciate it. That seems like a valuable feature if the data exists to incorporate into the assessment.

Cheers,

Thom

RE: Trouble with multiple growth morphs

2022-11-09T22:12:26Z

Hi Thom,

Good question. I think it only makes sense to use multiple growth morphs if you have separate data on each morph.

In a case where you have two areas with separate growth, then the indices and comp data for each area could separately inform the recruitment allocation that fits the data best in the same way that they would inform R0 in a single-area/single-morph model. Time-varying recruitment allocation may be harder to estimate well but in theory could be estimated similarly to recdevs in a single area.

In a case where two growth morphs overlap in space, you would need some other measure of the ratio between them which could be entered using the morph composition data type (described here: https://nmfs-stock-synthesis.github.io/doc/SS330_User_Manual.html#stock-morph-composition-data).

-Ian

RE: Trouble with multiple growth morphs

2022-11-09T21:57:23Z

Hello all,

When estimating recruitment apportionment for a multiple growth morph model, how does the data inform these parameters in SS?

Cheers,
Thom

RE: Forecast F scenarios by fleet

2022-10-26T07:16:23Z

Hi Ian,

Thanks a lot for the suggestion.

As you said, this option may be faster than finding the value iterating over 3 fleet F's.

After some trials I didn't manage to do it yet but I will keep exploring it and I also want to try something similar with the catches instead of F's (I think It can be a bit more intuitive).

Thanks,

Fran

RE: Forecast F scenarios by fleet

2022-10-19T16:42:40Z

Hi Fran,

Rick can correct me if I'm wrong, but I would think that you could use this option near the top of the forecast file

5 # Forecast: -1=none; 0=simple_1yr; 1=F(SPR); 2=F(MSY) 3=F(Btgt) or F0.1; 4=Ave F (uses first-last relF yrs); 5=input annual F scalar

(followed by input of the annual F scalar).

Then use these option further down in the file to turn on allocation among fleets:

2 # fleet relative F:  1=use first-last alloc year; 2=read seas, fleet, alloc list below

(followed by a table with the relative F for each fleet in each season, ending with a -9999 row).

I'm sorry I don't have time to try to work up an example right now to try it out, but you can keep checking the echoinput.sso file to see if the model is interpreting your inputs correctly.

It may take some iteration to get the annual F scalar correct, but iterating one value is better than iterating over the three F values for the fleet-specific F in each year.

-Ian

Forecast F scenarios by fleet

2022-10-19T13:47:29Z

Dear SS experts,

I am trying to do forecast scenarios in order to test the contribution to the SSB when reducing to F0 the trawl (medium-size individuals) and longline (big-size individuals) fleets.

I am using F report = 3 (sum apical F's by fleet) as it is the same F that I incude into the forecast file to generate the scenarios. One scenario is the base case, where apical F's from the average 2018-2020 period are projected in equilibrium for all fleets. The second scenario is the trawl fleet F0 case, where apical F is 0 only for trawl fleet and the third scenario would be the same with longline fleet.

I have tried by setting the same sum of apical F's by year in the different scenarios (by summing to the other fleets the F value rested to trawl, for instance) but the resulting scenarios display different annual F's, so I can not properly compare their SSB contribution (see attached plot). I guess that this input apical F is different from the resulting annual F as it must be influenced by selex, M and morphs.

I am probably missing something, but I wonder if there is any option or trick that allows to have the same annual F for these scenarios but having different fleet F contribution in each one.

Thanks a lot,

Fran

RE: Selectivity at size by sex

2022-08-31T18:56:50Z

Dear Rick,

Thank you for your quick response.

I had not realized about this option. As you mentioned, my selectivity was set as the same for males and females and that is why I found the same selex at size but different selex at age.

I will explore this option.

Thanks again,

Fran

RE: Selectivity at size by sex

2022-08-31T18:28:29Z

Dear Francisco,

Selectivity-at-length can differ by sex in SS3, but only if your setup for selectivity requests sex-specific selectivity. In the control file, it is the 3rd column of the selectivity setup and is labelled "male" and is described in the manual.

#_Pattern Discard Male Special
24 0 0 0 # 1 Fishery
24 0 0 0 # 2 Survey

Even if selectivity-at-length is the same for males and females, selectivity-at-age will differ between the sexes if growth is sex-specific. This is because selectivity-at-age is partly determined by the combination of selectivity-at-length and distribution of length-at-age.

Rick

RE: Selectivity at size by sex

2022-08-31T17:50:50Z

La cuenta de correo francisco.izquierdo@ieo.es dejará de estar disponible en breve. Para cualquier comunicación, por favor, envíe los mensajes a la nueva cuenta francisco.izquierdo@ieo.csic.es. Los correos enviados a francisco.izquierdo@ieo.es serán desviados a la nueva cuenta, por lo que no será necesario el envío a ambos buzones.

Selectivity at size by sex

2022-08-31T17:49:57Z

Dear SS experts,

I am working with a sex-separated length-based model for hake and I have a question regarding the report file output.

I have noticed that the majority of output tables by sex in the report file are provided at age and not at length (e.g. F, M or Catch). However, looking at the replist$sizeselex table I found the column "Sex". The thing is that both sexes (1 or 2) show exactly the same selectivity at length so I am wondering if this is an error or there is some way to get selex at length by sex.

I have read the section report in page 175 of the manual but I couldn't find the answer to this.

Many thanks in advance,

Fran

Bias when re-fitting data

2022-06-21T09:42:57Z

Dear SS3 team,

In order to work on an MSE (management strategy evaluation), I am doing the exercise of taking the output file (data_expval.ss) and using it as the input data file (naming it as ss3.dat). When I compare the original results to these "re-fitted" results, there is a quite high bias, up to a 20% for the las year SSB. I checked for convergence issues, and I did the jittering exercise, but it seems there is are no convergece problems. I also simulated some error to add to the data_expval.ss values, but the biased results remain similar.

Is this bias something expected or can it be related to the SS model configuration?

I attach a figure comparing results and the model files.

Thank you!

RE: Error about mean recruitment in forecast

2022-05-25T16:25:18Z

Hi Laura.

What you are seeing is a new warning regarding an illogical situation that could arise when using a relatively new feature: mean recruitment in forecast.

The warning contains the essential instruction. In your control file, change the phase for forecast recruitment to be negative, so not estimated:

-5 #_forecast_recruitment phase (incl. late recr) (0 value resets to maxphase+1)

Apology for fact that you invoke use of mean recruitment in the forecast file and that forces you to go to the control file to change the dev phase. That happens when we gradually add model features while maintaining backward compatibility of file structure.

Error about mean recruitment in forecast

2022-05-25T13:41:58Z

Hello all,

I am running projections and received an error I have not come across before: "mean recruitment for forecast is incompatible with pos. phase for forecast rec_devs; set phase to neg. unless using late rec_devs". I would appreciate any advice regarding this error and how to move forward.

Thanks in advance for your help!

Best fishes,

Laura

RE: Model scale problem

2022-05-12T06:34:52Z

Dear Rick,

Thank you for your explanation and advice on how to solve the problem. I applied them as described in the manual and I think they are very useful. However, the "F balpark" didn't work and regarding the "depletion survey", you have to be sure you know the ratio between B0 and SSB0 for that year as you somehow force SS to follow it.

After some trials I think that the problem on the scale may also come from the single area configuration. The dataset has been simulated for a spatial model but I am still fitting everything to a single area model first, so probably, when I increase the number of areas, the tag data will be more informative on the model scale as you mentioned.

Thanks also to Chantel for the suggestions although in this case I have not been the one to simulate the data, so I don't have all the information about quantities that I would need.

Regards,

Fran

RE: Error Message: matrix bound exceeded

2022-05-02T12:54:27Z

Thank you!

RE: Error Message: matrix bound exceeded

2022-05-02T12:50:45Z

Dear

The error is due to the weight at age file, you need a 1 and a 0 for the sex column. Then model runs but it gave weirds results so better you check it out more carefully.

Cheers

Max