Sampling-related and knots-related density covariates

[BioCatFish]

Hi Jim,

I wonder if there is a way to use sampling-related SST as a density covariate to fit the model but use knots-related SST to predict density and calculate total biomass?

What I'm trying to do is standardize the abundance of actively migrating juvenile salmon in the Sea of Okhotsk over multiple years. Surveys differ in temporal and spatial coverage. Usually, a survey takes up to 20-30 days to conduct. I use day-of-the-year as a catchability covariate to "control" the time difference in sampling within and between surveys.

As a density covariate, I plan to use SST, since juvenile salmon inhabit areas with certain temperature ranges (say 8-10 degrees). The problem is that during the time of the survey (20-30 days), SST throughout the sea cools down severely, and if I use any remote-sensing SST as "covariate_data" (for instance, monthly averaged or at a certain date), SST at many locations would differ from SST at the day of sampling and the relationship between encounter probability/catch rates and SST would be violated in some ways.

So, I would like to use sampling-related SST to set the model, but predict densities and estimate indexes using "standardized" SST (say, on the median day-of-the-year - for example, October 25th each year).

Hope my question makes sense to you.