R/process_bigeye_data.R

In peterkuriyama/tunast:

#' Function to process bigeye data
#' Process three types of bigeye data: annual counts, annual counts by length bins, 
#' annual counts by length bins and seasons
#' @param length_bins Vector of lengths to assign as 'species'
#' @export

process_bigeye_data <- function(length_bins){
  #-----------------------------------------------------------------------------------------------------
  #Read in Data
  bill_catch <- read.csv("data/PublicLLTunaBillfishMt.csv", stringsAsFactors = FALSE)
  bill_numbers <- read.csv("data/PublicLLTunaBillfishNum.csv", stringsAsFactors = FALSE)
  
  #Filter the billfish data to keep bigeye from Japan dat
  bet_catch <- bill_catch %>% filter(Flag == "JPN") %>% select(Year, Month, Flag,
    LatC5, LonC5, Hooks, BETn, BETmt)
  bet_numbers <- bill_numbers %>% filter(Flag == "JPN") %>% select(Year, Month, Flag,
    LatC5, LonC5, Hooks, BETn, BETmt)
  
  #Change column formatting
  bet_numbers <- plyr::rename(bet_numbers, c("LatC5" = "Lat", "LonC5" = "Lon"))
  names(bet_numbers) <- tolower(names(bet_numbers))
  
  #Calculate annual values
  bet_numbers_annual <- bet_numbers %>% group_by(year, lat, lon) %>% 
    summarize(hooks = sum(hooks), betn = sum(betn)) %>% as.data.frame
  
  bet_numbers_complete <- bet_numbers_annual %>% complete(year, nesting(lat, lon), 
    fill = list(hooks = 0, betn = 0)) 
  
  #Check proportion of zeroes
  bet_numbers_complete %>% group_by(year) %>% 
    summarize(zeroes = length(which(betn == 0)),
      nrowz = length(betn), prop_zero = zeroes / nrowz) %>% tail
  bet_numbers_complete$cpue <- bet_numbers_complete$betn
  bet_numbers_complete$spp <- 1
  bet_numbers_annual <- bet_numbers_complete %>% as.data.frame
  
  #---------------------------------
  #Composition data
  bet_comps <- read.csv("data/bet_length_comps.csv", stringsAsFactors = FALSE)
  
  #Classify the species
  for(ii in 1:length(length_bins)){
    # print(ii)    
    #Conditions of assigning species
    if(ii == 1) inds <- which(bet_comps$Bin <= length_bins[ii])  
    if(ii != 1) inds <- which(bet_comps$Bin > length_bins[ii - 1] & bet_comps$Bin <= length_bins[ii])
    
    # if(ii == length(tens)) inds <- which(bet_comps$Bin > tens[ii])
    
    bet_comps[inds, 'spp'] <- ii
    if(ii == length(length_bins)){
      inds <- which(bet_comps$Bin > length_bins[ii])
      bet_comps[inds, 'spp'] <- ii + 1
    }
  }
  
  #Treat certain length bins as species
  names(bet_comps) <- tolower(names(bet_comps))
  

  #Change frequency of all rows to 1 (because they are expanded)
  # bet_comps_expd$freq <- 1
  # sum(bet_comps_expd$freq) == sum(bet_comps$Freq)
  
  #Group on annual or seasonal time scales
  bet_comps_annual <- bet_comps %>% group_by(lat, lon, year,
    spp) %>% summarize(cpue = sum(freq)) %>% as.data.frame

  bet_comps_seasonal <- bet_comps %>% group_by(lat, lon, year, quarter, spp) %>%
    summarize(cpue = sum(freq)) %>% as.data.frame
  
  outs <- list(bet_numbers_annual = bet_numbers_annual, 
               bet_complete_annual = bet_comps_annual,
               bet_complete_seasonal = bet_comps_seasonal)
  
  return(outs)
}

  
  #Consider 'CPUE' to be numbers of fish
  #**might make this numbers/nhooks in the future**
  # bet_comps1 <- bet_comps %>% group_by(lat, lon, year, quarter, spp) %>% 
  #   summarize(cpue = sum(freq))
  
  #**Use annual "cpue" for now can also switch to include the quarter columns
  # bet_comps_annual <- bet_comps %>% group_by(lat, lon, year, spp) %>% 
  #   summarize(cpue = sum(freq)) %>% as.data.frame
  
  #**Also filter data so year >= 1990**
  #Just to get model to fit
  # bet_comps_annual <- bet_comps_annual %>% filter(year >= 1990)
  
  #Fill the missing values
  # bet_comps_annual
  
  #Overwrite old version with filled version
  # bet_comps_annual <- bet_complete_annual %>% filter(year >= 1986)
  # unique(bet_comps_annual$year)[order(unique(bet_comps_annual$year))]

###plot the data to see what things look like
  # world_map <- map_data("world")
  # plot_numbers <- bet_numbers_annual %>% group_by(lat, lon) %>% summarize(cpue = sum(cpue)) 
  
  # ggplot() + geom_raster(data = plot_numbers, aes(x = lon, y = lat, fill = cpue)) + 
  #   scale_fill_gradient(low = 'white', high = 'red') +
  #   geom_map(data = world_map, map = world_map,
  #     aes(x = long, y = lat, map_id = region)) + 
  #   scale_x_continuous(limits = c(-148, -72)) + 
  #   scale_y_continuous(limits = c(-43, 48)) + theme_bw()
    
  # bet_complete <- bet_co_annual %>% complete(spp, nesting(lat, lon, year),
  #   fill = list(cpue = 0)) %>% as.data.frame