Resolve how to fill in rec catch estimates in 1990-1992

[brianlangseth-NOAA]

MRFSS era estimates for 1990-1992 do not exist because sampling did not occur. We will need to decide how to resolve. For 2021 assessment used linear interpolation between 1989 and 1993. Do we want to apply the same approach?

[brianlangseth-NOAA]

Was there sampling of the PC fleet in 1993-1995? Fleet specific landings in MRFSS were minimal (though perhaps not all that different with 1980 and 1988).

[brianlangseth-NOAA]

For canary rockfish (see issue pfmc-assessments/canary_2023#50), mortality is less variable year to year, and the 1993-1995 time period was not near the highest estimates. For canary, which is presumably better sampled, the low PC estimates were obvious, and fixing was more straightforward. For quillback however, the sampling is intermittent. Thus, adding presumed PC estimate to 1993 makes it the single greatest year of sampling, which would seem like an overestimate. Adding PC amounts to 1994-1995, which are already small, would not greatly affect results.

Im inclined not to make the same change for quillback. U

UPDATE: See comment below. Looks like there was a true gap is sampling for these years so adding would make sense.

[brianlangseth-NOAA]

For filling in 1990-1992, the choice of option has a decently large effect. The blue points impute between 1989 and 1993, while the red line is the average of nearby estimates (1987-1989 for 1990, 1987-1989 and 1993-1995 for 1991, and 1993-1995 for 1992). If the assumption from the 2021 assessment was used (average of 1989 and 1993), the values from 1990-1992 would be the same as the middle blue dot (black line). Plan to use the red line

To summarize:
-Red line: moving average like. Average previous three years, the three years on each side, then the following three years.
-Blue dots: interpolation between previous (1989) and following year (1993)
-Black line: taking the average of the previous and following years (1989 and 1993)

[brianlangseth-NOAA]

With respect to 1993-1995, Melissa looked at historical patterns and found that sampling restarted in the north (north of san luis obispo) in 1996, but restarting in 1993 in the south. Thus, adding PC mode catch to the estimates of PR would seem to make sense.

[brianlangseth-NOAA]

Ratios of PC:PR are shown below, with zeros for 1980 and 1988 not yet filled in. The average of all years (except 1993-1995) is 0.7468163. Adding this to the PR estimates for 1993-1995 is a way to account for unsampled PC estimates.

Based on this, the corresponding extra amount of PC would be

[brianlangseth-NOAA]

Commit 2c76e36 made final changes for 1993-1995 (and correspondingly 1990-1992), which were based on taking an overall average of PC mortality from other years with data between 1980-2004. The intent was to minimize linking the expansion of catch from sparse data in one sector to another (which the ratio would do). See discussion in pfmc-assessments/california-data#2. Here is what this looks like: