Page 1 of 8 123 ... LastLast
Results 1 to 20 of 145

Thread: update on Kenai River sockeye models and escapement

  1. #1
    Member
    Join Date
    Jul 2006
    Posts
    5,533

    Default update on Kenai River sockeye models and escapement

    I thought I would start a new thread and one with data not speculation. Below is the ADF&G update on the Kenai River production models using the data from this year return. As you can read for yourself the Brood Year Interaction model is still the best fit and the Markov table has held firm. So the conjecture on needing a new model has not been supported. For escapements in the 600,000 to 900,000 range the mean yield is around 4 million while at greater than 1.3 million it is 2.8 million. The Markov table is what has happened in the past and is not a model per se. It is empirical data.


    Also the new projection on inriver escapement at the sonar counter this year will be around 1.7 million.

    Next year could change things again on model fit depending on how the return comes in.


    With the large Kenai return this year folks have been talking about whether the brood interaction model is history or if the EG is going to change. Assuming the final 2011 Kenai run is 5.7 million, I updated the brood table and estimated the fit of the various models we included in our 2010 EG report. The following is a brief comparison:

    2010 report Updated Model
    Ricker Model – R2 0.528 0.547
    BYI Model – R2 0.600 0.593
    I also updated the Markov table and the BYI simulation model (see below).
    So, this analysis indicated a small increase in the fit of the Ricker model and a small decrease in the fit of the BYI model, but the BYI model is still the best fit model. In the Markov table, the mean yield for escapements >1,300,000 increased from 2,566,000 to 2,863,000, but the highest mean yields were still in the 600,000 – 900,000 range. The BYI simulation with the updated model parameters indicated the goal range would change to 700,000 to 1,300,000. If our smolt data is accurate, we may have another large return of 2.3’s next year coming off the 2006 brood year.

  2. #2

    Default

    The model predicted that large escapements would reduce returns. We had large escapements. Returns did not decline.

    Model predictions have utterly failed to fit data from years after it was originally derived including this year and last. It doesn't matter if it seems to explain what happened before 1990 if it doesn't explain what has happened since. I don't care that rain used to be predicted by the price of tea in China if you can't tell me whether it is going to rain on me today.

    And the model still has all the same statistical construct problems it has always had (multiplicative interaction term with no main effect) - you remember, the ones identified by the deparment's cheif scientists.

  3. #3

    Default

    One question I have, how many years of data was analyzed and used for the development of the Brood Year Interaction Model?

  4. #4
    Member
    Join Date
    Jul 2006
    Posts
    5,533

    Default

    Quote Originally Posted by Bfish View Post
    The model predicted that large escapements would reduce returns. We had large escapements. Returns did not decline.

    And your model still has all the same statistical construct problems it has always had (multiplicative interaction term with no main effect.)
    Again you misquote and say it is my model. It has been refined since I worked at ADF&G and it is anADF&G model. In point of fact Bfish when I was part of the team that came up with the original model I advocated for variable escapement goals based on fry data. I lost that debate since management needed something the public could understand which was fixed goal ranges.

    Also, the simulation using the model predicted lower average returns from large escapements and that has happened - per the Markov table. So the data speaks for itself and the multiplicative interaction term has been peer reviewed and accepted by a number of statistical experts including ADF&G personnel would developed the terms. So we are back to not looking at the data but trying to discredit something based on what? Nothing but conjecture that did not hold up. We know your and KRSA position. What we need are some data to support it and that has not happened.

  5. #5
    Member
    Join Date
    Dec 2006
    Posts
    2,883

    Default

    Thanks Nerka. The BYI model does appear to be best fit. And it has sustained the run for decades.

  6. #6

    Default

    I'd be happy to loan you boys a shovel if you want to keep digging your holes.

  7. #7
    New member
    Join Date
    Feb 2004
    Location
    Soldotna
    Posts
    5,639

    Default Bingo . . . !!

    Quote Originally Posted by Bfish View Post
    The model predicted that large escapements would reduce returns. We had large escapements. Returns did not decline.

    Model predictions have utterly failed to fit data from years after it was originally derived including this year and last. It doesn't matter if it seems to explain what happened before 1990 if it doesn't explain what has happened since. I don't care that rain used to be predicted by the price of tea in China if you can't tell me whether it is going to rain on me today.

    And the model still has all the same statistical construct problems it has always had (multiplicative interaction term with no main effect) - you remember, the ones identified by the deparment's cheif scientists.
    At last! Something in plain, straightforward English. Something understandable.

    Thanks . .

  8. #8
    Member
    Join Date
    Dec 2006
    Posts
    2,883

    Default

    No need Bfish. All we need from you are the details of your, supposidly better, model...the supporting data, evidence, etc. Anyone can understand criticism and conjecture, even in plain straight forward english.

  9. #9
    Member
    Join Date
    Jul 2006
    Posts
    5,533

    Default

    Quote Originally Posted by Marcus View Post
    At last! Something in plain, straightforward English. Something understandable.

    Thanks . .
    You could not understand plain english if it was right in front of you. The quote from ADF&G is very plain and straight forward but your vindicative nature is coming out loud and clear on this one. You just cannot admitt it and now try to steal this thread. I started this thread to discuss data not some simplistic mind set from you. If you have data please provide it otherwise post on the thread you started.

  10. #10
    Member
    Join Date
    Jul 2006
    Posts
    5,533

    Default

    Quote Originally Posted by NorcalBob View Post
    One question I have, how many years of data was analyzed and used for the development of the Brood Year Interaction Model?
    The brood year interaction fry data started in 1985 or 86 and has been collected every year since. The adult data started in 1966.

  11. #11

    Default

    Quote Originally Posted by Nerka View Post
    I started this thread to discuss data not some simplistic mind set from you. If you have data please provide it otherwise post on the thread you started.
    Nerk, I'm not discounting the results of MW's quick and dirty analysis. It is new information and interesting - thanks for posting.
    At the same time I just see results and conclusions - no data or methods. If I posted something similar you and your grumpy buddy would rip me a new one.

    r-squares, markov tables and all that other science stuff aside, at the end of the day you are going to have to explain why your model appears to have no predictive power whatsover.

    So for now I'll be happy to look forward to the Department's updated escapement goal analysis in a couple of years.

    Wanna make a bet? (I'm giving 4 to 1 against on this one.)

  12. #12
    New member
    Join Date
    Feb 2004
    Location
    Soldotna
    Posts
    5,639

    Thumbs down

    You could not understand plain english if it was right in front of you. . . your vindicative nature is coming out loud and clear on this one. You . . now try to steal this thread. . . some simplistic mind set from you. . .
    Ad hominem abuse and personal slander only make civil discussion impossible.

  13. #13
    New member
    Join Date
    Feb 2004
    Location
    Soldotna
    Posts
    5,639

    Default

    Quote Originally Posted by Bfish View Post
    . . at the end of the day you are going to have to explain why your model appears to have no predictive power whatsover. . .
    An excellent point, Bfish, and one we all need to keep foremost in mind. If a scientific theory or model cannot predict future results, it isn't science, it's a failed theory/model.

    That's why double-blind studies are done . . in order to test the validity of a theory/model.


    A scientific model seeks to represent empirical objects, phenomena, and physical processes in a logical and objective way. All models are in simulacra, that is, simplified reflections of reality, but, despite their inherent falsity, they are nevertheless extremely useful.[2] Building and disputing models is fundamental to the scientific enterprise. Complete and true representation may be impossible (see non-representational theory), but scientific debate often concerns which is the better model for a given task, e.g., which is the more accurate climate model for seasonal forecasting.[3]

    Attempts to formalize the principles of the empirical sciences, use an interpretation to model reality, in the same way logicians axiomatize the principles of logic. The aim of these attempts is to construct a formal system that will not produce theoretical consequences that are contrary to what is found in reality. Predictions or other statements drawn from such a formal system mirror or map the real world only insofar as these scientific models are true.[4]

    For the scientist, a model is also a way in which the human thought processes can be amplified.[5] For instance, models that are rendered in software allow scientists to leverage computational power to simulate, visualize, manipulate and gain intuition about the entity, phenomenon, or process being represented. Such computer models are in silico. Other types of scientific model are in vivo (living models, such as laboratory rats) and in vitro (in glassware, such as tissue culture).
    [6]


  14. #14
    Member
    Join Date
    Dec 2006
    Posts
    2,883

    Default

    Bfish, the results and conclusions from Nerka are supported by a well-known and well-documented method, along with years of documented data. Supporting evidence itself can be found in the resulting years of a sustained run. And that is where your ideals fall short...They introduce nothing but criticism, speculation, and conjecture.

    Also, no honest model would attempt to have the power of predicting the future (except yours-whatever it is). The BYI model uses the best data available to make the best predictions it can, while recognizing uncertainty, and in fact allowing for it. That is why the goal has a range, why it can be moved, and why it takes into account uncertainty. So no, it is you, critic of the best fit model, who at the end of the day must explain your model - whatever it is. Again, enough conjecture.

    BTW, calling me names and choosing my buddies is unbecoming of you.

  15. #15
    New member
    Join Date
    Feb 2004
    Location
    Soldotna
    Posts
    5,639

    Question Scientific models . . .

    Bfish, . . no honest model would attempt to have the power of predicting the future . .


    —from a couple Internet sites that came up when I Googled "scientific model":

    If a "scientific model" cannot predict "the outcome of related situations," it is a failed model.


    what is a scientific model?

    . . a scientific model is a “testable idea… created by the human mind that tells a story about what happens in nature.” another definition is “a description of nature that can predict things about many similar situations.” models are developed when a scientist’s creativity and insight are combined with data and observations about many similar scenarios. Scientists try to identify and generalize patterns in these observations, and use mathematical language to predict the outcome of related situations. The value of a model is that we can trust its predictions about similar situations even if we don’t encounter each situation.
    models and objective truth: the qualitative and quantitative models are part of man’s attempt to study nature and discover the underlying laws. However, the models do not represent the final truth.

    all models are only approximations. They are proposed at the simple and simplified level, and then compared with the actual phenomenon. The model is then reject, modified, or refined based upon feedback. This way the model moves from a crude approximation to a highly accurate and complex mathematical construct.

    the process of defining and then refining a model may go on for decades or even more. For example the current model of the atom is almost a century old, but refinements are still going on and the final model is nowhere in sight. What’s more, the discovery in the nineteen sixties of sub atomic particles known as quarks have shown that the structure of the atoms is incredibly more complex then what anyone could ever imagine, and that perhaps another century of additions and refinements will be needed to arrive art a comprehensive models of the atom.

  16. #16
    Member
    Join Date
    Dec 2006
    Posts
    2,883

    Default

    Marcus, I hate to break it to you (again), but this discussion is not about what a scientific model is, your opinions on ad hominem abuse, or how understandable you think someone's post was. You appear to be digressing the topic of Nerka's update on the Kenai River sockeye models to your usual diatribe. How about you nip it in the bud right now, and maybe we can actually accomplish something here....

    By the way, you left our last conversation, "Fresh Regulatory Ideas for the Kenai" without answering my simple question. The thread is still open, and I would appreciate an answer.

  17. #17
    New member
    Join Date
    Feb 2004
    Location
    Soldotna
    Posts
    5,639

    Default Defining the terms of the discussion . .

    Marcus, . . this discussion is not about what a scientific model is, . .
    . . . . . . . .

    Quote Originally Posted by Nerka View Post
    . . Below is the ADF&G update on the Kenai River production models using the data from this year return. As you can read for yourself the Brood Year Interaction model is still the best fit. . the conjecture on needing a new model has not been supported. . . The Markov table is what has happened in the past and is not a model per se. It is empirical data. .

    Next year could change things again on model fit depending on how the return comes in.

    With the large Kenai return this year folks have been talking about whether the brood interaction
    model is history or if the EG is going to change. Assuming the final 2011 Kenai run is 5.7 million, I updated the brood table and estimated the fit of the various models we included in our 2010 EG report. The following is a brief comparison:
    2010 report Updated Model
    Ricker Model – R2 0.528 0.547
    BYI Model – R2 0.600 0.593
    I also updated the Markov table and the BYI simulation model (see below).
    So, this analysis indicated a small increase in the fit of the Ricker model and a small decrease in the fit of the BYI model, but the BYI model is still the best fit model. In the Markov table, the mean yield for escapements >1,300,000 increased from 2,566,000 to 2,863,000, but the highest mean yields were still in the 600,000 – 900,000 range. The BYI simulation with the updated model parameters indicated the goal range would change to 700,000 to 1,300,000. If our smolt data is accurate, we may have another large return of 2.3’s next year coming off the 2006 brood year.

  18. #18
    Member
    Join Date
    Jul 2006
    Posts
    5,533

    Default

    The models in question have been around for a long time. The Ricker model was tried for years but the problem with it was that it was never statistically significant. Thus ADF&G started to look for something better and came up with the Brood year interaction model. Both models have been tested since 1999 with each new year of data and the Brood year interaction model has held up as the best fit and the Ricker has improved but still not the best fit to the data.

    Neither model is used to predict a given return from a given brood year escapement. Forecast of UCI returns are done with fry data, age composition data - for example the relationship of 4 year old to 5 year old fish. The two models in question are being used to help evaluate escapement goals that includes limnological data from the lake, fish size relative to density, and other parameters. Thus the escapement goal team is more comprehensive.

    Marcus post is true but he misses the point of annual prediction vs long term averages. The simulation model I referenced shows that in some years high escapements will produce large returns. It is all a function of fry numbers in the lake and the interaction of the brood years relative to survival. Not understanding the data Marcus has made the mistake many make in this discussion. One year of results do not refute a model of this nature.

    Just again for the record - the actual returns from know escapements show that about 4 million fish are produced from 600,000 to 900,000 spawners and only about 2.8 million from spawners greater than 1.3 million. These numbers include this years return. So on average there is 1.2 million fish less in the harvest on average if one goes with higher escapements. These data are collected over 45 years so the data set does include lots of variation which is good to see how the system behaves over time.

    Finally, major changes in the freshwater or marine environment could make both models insignificant with time. That has not happened so saying that it may happen is not very meaningful to the discussion today. Bfish wants to bet that one model will get poorer and the other better in the next two years. Maybe but that is not science it is conjecture. However, if he is advocating a change right now based on that conjecture then it is not science.

    Relative to Marcus and the idea of a failed model there is a saying in science - most scientific directions fail and that leads to success in understanding. So failed is not a negative in science. It can lead to great advancements in our understanding of the natural world.

    Finally, I stated Marcus could not understand plain English as an example of how his comment about plain English was a slander on all who posted on this thread other than him and Bfish I guess. It has no place here and I agree with that but Marcus needs to see that his comment is personal when he states all others do not speak in plain English and he only makes that reference with someone he agrees with. No more comments like that from me and I hope Marcus would do the same.

  19. #19
    New member
    Join Date
    Feb 2004
    Location
    Soldotna
    Posts
    5,639

    Default

    . . Bfish wants to bet that one model will get poorer and the other better in the next two years. Maybe but that is not science it is conjecture. However, if he is advocating a change right now based on that conjecture then it is not science. . .
    Bfish? . . . . .

  20. #20
    Member
    Join Date
    Dec 2006
    Posts
    2,883

    Default

    Very good Nerka. I'm sure if another model can show to be a better fit on down the road, we will welcome it. After all, we want the best for our fisheries. Until then it is speculation and conjecture.

    We give to much attention to some here. When one does not contribute anything but antagonizing digression in thread after thread, it is hardly worth engaging. When one has to use ridiculous wordsmithing to cover his tracks, the diatribe is sure to follow. Don't feed the antagonizing trolls.

Page 1 of 8 123 ... LastLast

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •