In Craig44/stockassessmenthelper: Package to help plot stock assessment models and functionality

Catch and Effort grooming code {#CatchEffort}

This section shows functions and methods that I have employed for analysising and grooming catch and effort fishery data.

Grooming and generating a core data set

Fishery-dependent catch and effort data generally contain reporting and typographic errors such as fishing events on land or unrealistically large catches. These fishing events are identified and removed to avoid influence in any analysis, a process often coined "grooming" in the CPUE standardisation literature [@bentley2012influence]. Another consideration of fishery-dependent catch and effort data is it often contains a number of events which occur outside the main stock habitat i.e., outside known depth range. In addition, some vessels may only operate in the fishery for short periods of time or sporadically over time. Information from these fishing events are unlikely to inform changes in relative abundance over the long term. A set of rules are generally applied to develop a "core" data set which represents consistent fishing behavior over time, space and with respect to fishing method.

A key component when creating a "core" data set is to demonstrate what the effect is on dropping non core fishing events. This includes how much catch, how many records, location and if the records dropped have common covariates that is separate to the core data set. Functions available for documenting grooming and subsetting rules include record_grooming_rule and apply_grooming_rule which are shown below.

catch_effort_data = readRDS(file = system.file("extdata", "simulated_data.RDS", 
                                               package = "stockassessmenthelper", mustWork = TRUE))
# look at the data
# head(catch_effort_data)

## start grooming record
start_records = nrow(catch_effort_data)
## these are records to save the effects of grooming rules
start_ndx = rep(T, nrow(catch_effort_data))
total_grooming_record = record_grooming_rule(df = catch_effort_data, index = start_ndx, 
                                             catch.col = "catch", record = NULL, rule = "Init") 
yearly_catch_record = record_grooming_rule(df = catch_effort_data, index = start_ndx, catch.col = "catch", 
            year.col = "fish_year", record = NULL, rule = "Init", attribute = "cumulative_catch")
yearly_event_record = record_grooming_rule(df = catch_effort_data, index = start_ndx, catch.col = "catch", 
            year.col = "fish_year", record = NULL, rule = "Init", attribute = "cumulative_events")


## Remove unrealistic large catches (possible should consider imputing)
large_catch_ndx = catch_effort_data$catch < 40000
# record the rule
total_grooming_record = record_grooming_rule(df = catch_effort_data, index = large_catch_ndx, 
                      catch.col = "catch", record = total_grooming_record, rule = "catch > 40 t") 
yearly_catch_record = record_grooming_rule(df = catch_effort_data, index = large_catch_ndx, 
                      catch.col = "catch", year.col = "fish_year", record = yearly_catch_record, 
                      rule = "catch > 40 t", attribute = "cumulative_catch")
yearly_event_record = record_grooming_rule(df = catch_effort_data, index = large_catch_ndx, 
                      catch.col = "catch", year.col = "fish_year", record = yearly_event_record, 
                      rule = "catch > 40 t", attribute = "cumulative_events")
# apply the rule to the data set
catch_effort_data = apply_grooming_rule(catch_effort_data, large_catch_ndx)

## Only look at Bottom Trawl data
method_BT_ndx =catch_effort_data$method == "BT"
# record the rule
total_grooming_record = record_grooming_rule(df = catch_effort_data, index = method_BT_ndx, 
                      catch.col = "catch", record = total_grooming_record, rule = "method BT") 
yearly_catch_record = record_grooming_rule(df = catch_effort_data, index = method_BT_ndx, 
                      catch.col = "catch", year.col = "fish_year", record = yearly_catch_record, 
                      rule = "method BT", attribute = "cumulative_catch")
yearly_event_record = record_grooming_rule(df = catch_effort_data, index = method_BT_ndx, 
                      catch.col = "catch", year.col = "fish_year", record = yearly_event_record, 
                      rule = "method BT", attribute = "cumulative_events")
# apply the rule to the data set
catch_effort_data = apply_grooming_rule(catch_effort_data, method_BT_ndx)

If you apply the above functions you end up with records that can be plotted up using ggplot e.g

## plot records
melt_total = melt(total_grooming_record, id.vars = "rule")
melt_total$rule = factor(melt_total$rule, ordered = T, levels = total_grooming_record$rule)
ggplot(melt_total %>% filter(variable %in%  c("catch")), aes(y = value /1000, x = rule, group = 1)) +
  geom_line(size = 2, linetype = "dotted") +
  geom_point(size = 4, aes(col = rule)) + 
  theme(axis.text.x = element_text(angle = 90)) +
  ylab("Catch (000's t)") +
  ylim(0,NA)

ggplot(melt_total %>% filter(variable %in%  c("events")), aes(y = value /1000, x = rule, group = 1)) +
  geom_line(size = 2, linetype = "dotted") +
  geom_point(size = 4, aes(col = rule)) + 
  theme(axis.text.x = element_text(angle = 90)) +
  ylab("events (000's)") +
  xlab("") +
  ylim(0,NA)

Craig44/stockassessmenthelper documentation built on April 14, 2023, 10:57 a.m.