Unless otherwise noted all North Coast chum data provided by Shaun Davies and Katie Beach (DFO)
NBBR/Age Tables/CM OUTPUT AGE StatArea.xlsx
MultiStockHierPSF.txt
which was replicated in our notation as MultiStockHierPSFRep.txt
CU prSmax prCV
Skeena_Estuary 1118 1
Lower_Skeena 20378 10
Middle_Skeena 2544 1
Portland_C_Obs 42230 10
Portland_Inlet 24524 10
ncStockRecModels.R
; some details below in Apr. 30 updateFirst pass on fitting SR models - Feb. 13
ncStockRecModels.R
and stockRecFunctions.R
stockRecFunctions.R
escapement time series used to calc sMax, while paired escapement/recruitment used to estimate SRif(stk==1){ CU<-"Portland Inlet" }; if(stk==2){CU<-"Lower Nass" };
if(stk==3){ CU<-"Portland Canal-Observatory" }
Validate age structure (RESOLVED) - Mar. 8
for(i in 1:nrow(newChumDat)){
age3Esc <- newChumDat$Age3[i+3]*newChumDat$Escape[i+3]
age3Catch <- newChumDat$Age3[i+3]*(newChumDat$Total.ER[i+3]*(newChumDat$Escape[i+3]/(1-newChumDat$Total.ER[i+3])))
newChumDat$rec3[i] <- age3Catch+age3Esc
age4Esc <- newChumDat$Age4[i+4]*newChumDat$Escape[i+4]
age4Catch <- newChumDat$Age4[i+4]*(newChumDat$Total.ER[i+4]*(newChumDat$Escape[i+4]/(1-newChumDat$Total.ER[i+4])))
newChumDat$rec4[i] <- age4Catch+age4Esc
age5Esc <- newChumDat$Age5[i+5]*newChumDat$Escape[i+5]
age5Catch <- newChumDat$Age5[i+5]*(newChumDat$Total.ER[i+5]*(newChumDat$Escape[i+5]/(1-newChumDat$Total.ER[i+5])))
newChumDat$rec5[i] <- age5Catch+age5Esc
age6Esc <- newChumDat$Age6[i+6]*newChumDat$Escape[i+6]
age6Catch <- newChumDat$Age6[i+6]*(newChumDat$Total.ER[i+6]*(newChumDat$Escape[i+6]/(1-newChumDat$Total.ER[i+6])))
newChumDat$rec6[i] <- age6Catch+age6Esc
}
Adjustments to stock recruit models - Mar. 16
MultiStockHierCF.txt
nassChumMCMCPars.csv
contains alpha values from the posterior hyperdistribution for alpha, beta0 value is equal to median from escapement time series and sigma is shared with Portland Canal-Observatory (larger sigma Portland Inlet)Discrepancies with PSF estimates - Apr. 30 (RESOLVED MAY 31)
MultiStockHierSimple.txt
modelMultiStockHierSimple.txt
modelMultiStockHierPSFRep.txt
modelMultiStockHierPSF.txt
model (basically a copy of B. Connors model with very minor changesdata/northCoastDat/psfMCMCPars.out
MultiStockHierSimple.txt
model with our own priorsIssues w/ data limited productivity - June 1
Concerns around forward simulated variability - June 5
May also be influenced by:
Catch comparison and outcome uncertainty - June 7
miscSensistivity/catchComparison/catchTrends.pdf
and analysis in misc/ncCatchComparison.R
simulationRunNotes.md
Update stock recruitment models - July 23 - Given evidence of autocorrelation in residuals, decided to re-estimate SR parameters using an AR Ricker model (MultiStockHierAR.txt) - Prior for rho ~ dnorm(0, 1) - Resulted in realistic estimates for all parameters - Increasing sd in prior to 10 resulted AR effects completely disappearing - Limited examples to compare to, but generally analysts seem to use uninformative prior w/ broad uniform distribution - At same time changed prior on hyperdistribution for alpha to be set on TauA_global ~ dgamma(1, 1) - I.e. sigA_global no longer defined - Data limited CU's parameters estimated by drawing 1000 samples from dataset that contains the other 5; beta based on 1/median observed escapement
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.