# Define planning regions and choose which species count as important to fisheries


###########################
## Define planning regions
###########################
require(data.table)
require(sf)

gridsz <- 0.25 # grid size for the CMSP analysis

# read in data
grid <- readRDS('temp/SPsf2.rds') # analysis grid
eez <- st_read('dataDL/marineregions/World_EEZ_v10_20180221/eez_v10.shp')
coast <- st_read('dataDL/natcap/Marine/Land/global_polyline.shp') # useful for checking success by making maps

# get ready to define regions
indsCAN <- st_intersects(eez[eez$Pol_type == '200NM' & eez$Territory1 == 'Canada',], grid, sparse = FALSE) # slow...
grid$region <- NA

# define each region (order is important, as later ones build off earlier ones)
# NEUS
grid$region[grid$lat > 35.6 & grid$lat < 45 & grid$lon > - 100 & grid$lon < -60 & !indsCAN] <- 'neus'

# Newfoundland
grid$region[(indsCAN & grid$lon > -100 & (grid$lon > -53 | (grid$lat > 49 & grid$lon > -57) | (grid$lat > 52) | (grid$lat > 47.8 & grid$lon > -54) | (grid$lat > 47 & grid$lon > -53.5))) | (grid$lon > -53 & grid$lon < -40)] <- 'newf'

# maritime Canada (including French waters around St. Pierre & Miquelon)
grid$region[(indsCAN & is.na(grid$region) & grid$lon > -100) | (grid$lat > 44 & grid$lat < 48 & grid$lon > -57.2 & grid$lon < -55)] <- 'maritime'

# SEUS
grid$region[(grid$lon > -80.6 | (grid$lon > -82.3 & grid$lat > 28)) & grid$lat <= 35.6] <- 'seus'

# GoMex
grid$region[grid$lon > -100 & grid$lat < 31 & is.na(grid$region)] <- 'gmex'

# west coast US
grid$region[grid$lon < -100 & grid$lat < 50 & !indsCAN] <- 'wc'

# British Columbia
grid$region[grid$lon < -100 & indsCAN] <- 'bc'

# GoAK
grid$region[grid$lon < -100 & grid$lon > -156 & is.na(grid$region)] <- 'goa'

# EBS and Aleutians
grid$region[grid$lon <= -156 | grid$lon > 0] <- 'ebs'

# any missing?
sum(is.na(grid$region))
    with(grid[is.na(grid$region), ], plot(lon, lat, cex=0.1)) # plot the missing points

# summarize into CMSP analysis grids
grid$latgrid <- floor(grid$lat/gridsz)*gridsz + gridsz/2
grid$longrid <- floor(grid$lon/gridsz)*gridsz + gridsz/2
gridcmsp <- as.data.table(grid)[, .(region = names(sort(table(region), decreasing = TRUE))[1], nregion = length(unique(region))), by = c('latgrid', 'longrid')] # take the most common region in each grid cell

# plot
gridcmsp[region == 'newf', plot(longrid, latgrid, cex=0.1)]
gridcmsp[region == 'maritime', plot(longrid, latgrid, cex=0.1)]
gridcmsp[region == 'neus', plot(longrid, latgrid, cex=0.1)]
gridcmsp[region == 'seus', plot(longrid, latgrid, cex=0.1)]
gridcmsp[region == 'gmex', plot(longrid, latgrid, cex=0.1)]
gridcmsp[region == 'wc', plot(longrid, latgrid, cex=0.1)]
gridcmsp[region == 'bc', plot(longrid, latgrid, cex=0.1)]
gridcmsp[region == 'goa', plot(longrid, latgrid, cex=0.1)] # odd chunk out of SW corner? but nearly all is land
    # plot(st_geometry(coast), add=TRUE) # bit slow
gridcmsp[region == 'ebs', plot(longrid, latgrid, cex=0.1)]

# write out
write.csv(gridcmsp, gzfile('output/region_grid.csv.gz'))

# gridcmsp <- fread('gunzip -c output/region_grid.csv.gz')

###########################################################
# Determine which species are of fishery importance
# from Sea Around Us fishery catch data
###########################################################
require(data.table)

## get list of projected species
files <- list.files(path = '/local/shared/pinsky_lab/projections_PlosOne2018/CEmodels_proj_Biomass_BCODMO', full.names = FALSE) # find relevant projection files
projspps <- unique(gsub(paste0('/|_Atl|_Pac|_rcp26|_rcp85|_jas_prediction_AGG.RData'), '', files))

# get tables of landings by LME from Sea Around Us
ebs <- fread('dataDL/sau/SAU LME 1 v47-1.csv')
ai <- fread('dataDL/sau/SAU LME 65 v47-1.csv') # will combine with ebs
goa <- fread('dataDL/sau/SAU LME 2 v47-1.csv') # shared with goa and bc
bc <- goa
wc <- fread('dataDL/sau/SAU LME 3 v47-1.csv')
gmex <- fread('dataDL/sau/SAU LME 5 v47-1.csv')
seus <- fread('dataDL/sau/SAU LME 6 v47-1.csv')
neus <- fread('dataDL/sau/SAU LME 7 v47-1.csv')
maritime <- fread('dataDL/sau/SAU LME 8 v47-1.csv')
newf <- fread('dataDL/sau/SAU LME 9 v47-1.csv')

# add region
ebs$region <- 'ebs'
ai$region <- 'ebs'
goa$region <- 'goa'
bc$region <- 'bc'
wc$region <- 'wc'
gmex$region <- 'gmex'
seus$region <- 'seus'
neus$region <- 'neus'
maritime$region <- 'maritime'
newf$region <- 'newf'

# combine and aggregate recent reported commercial landings by species
sau <- rbind(ebs, ai, goa, bc, wc, gmex, seus, neus, maritime, newf)
sppston <- sau[year >= 1995 & fishing_sector == 'Industrial' & catch_type == 'Landings' & reporting_status == 'Reported', .(tonnes = sum(tonnes)), by = c('region', 'scientific_name')]

# turn SAU names to lower case to match projections
sppston[, scientific_name := tolower(scientific_name)]

# set up vector hold names that match projections
sppston[, projname := as.character(NA)]

# match names from SAU to names from projections
for (j in 1:nrow(sppston)){
	ind <- agrep(sppston[j, scientific_name], projspps) # find index into spps
	if (length(ind) == 1) sppston[j, projname := projspps[ind]] # only enter if exactly one match
}

# order by decreasing landings
setkey(sppston, region, tonnes)

# manually fix any that agrep missed or filled in mistakenly (checked by eye)
sppston[scientific_name == 'clupea pallasii pallasii', projname := 'clupea pallasii']
sppston[scientific_name == 'theragra chalcogramma', projname := 'gadus chalcogrammus']
sppston[scientific_name == 'farfantepenaeus duorarum', projname := 'penaeus duorarum']
sppston[scientific_name == 'litopenaeus setiferus', projname := 'penaeus setiferus']
sppston[scientific_name == 'farfantepenaeus aztecus', projname := 'penaeus aztecus']
sppston[scientific_name == 'doryteuthis pealeii', projname := 'doryteuthis pealeii'] # agrep also gets doryteuthis pleii
sppston[scientific_name == 'stomolophus', projname := NA] # agrep mistakenly pulls a species name
sppston[scientific_name == 'harengula', projname := NA] # agrep mistakenly pulls a species name


# rank the remaining species and remove those that aren't in the projections
sppstontrim <- sppston[!is.na(projname), .(sau_name = scientific_name, projname = projname, tonnes = tonnes, rank = rank(tonnes)), by = 'region']


# trim to top 10 that are in projections for each region
sppstontrim[, maxrank := max(rank), by='region']
sppstontrim <- sppstontrim[(maxrank - rank) < 10, ]
sppstontrim[ , c('rank', 'maxrank') := NULL] # remove unneeded columns

# check that it worked
sppstontrim[, .N, by='region']

# write out
write.csv(sppstontrim, file='output/fishery_spps.csv')