Full Model Non-Convergence

[djhocking]

I get no mixing and non-convergence when I try running the full model with lots of lagged temperature and precipitation interactions. Moving forward, the question remains what variables/interactions to remove and whether we want to replace the lagged effects with moving means?

fixed.ef <- c("intercept" 
              , "Latitude" 
              , "Longitude" 
              , "TotDASqKM" 
              , "Forest" 
              , "ReachElevationM" 
              , "SurficialCoarseC" 
              , "CONUSWetland" 
              , "ImpoundmentsAllSqKM" 
              , "airTemp.TotDASqKM"
              , "airTempLagged1.TotDASqKM"
              , "airTempLagged2.TotDASqKM"
              , "prcp.Forest"
              )

site.ef <- c( "intercept.site" 
              , "airTemp" 
              , "airTempLagged1"
              , "airTempLagged2" 
              , "prcp" 
              , "prcpLagged1" 
              , "prcpLagged2"
              , "airTemp.prcp" # add combos of air*prcp or do longer moving means????
              , "airTemp.prcpLagged1"
              , "airTemp.prcpLagged2"
              , "airTempLagged1.prcpLagged1"
              , "airTempLagged1.prcpLagged2"
              , "airTempLagged2.prcpLagged2"
              )

year.ef <- c( "intercept.year"
              , "dOY" 
              , "dOY2"
              , "dOY3"
              )