In BigelowLab/ohwobpg: ohwobpg for Ocean Biology Processing Group data

Deriving rasters from other rasters

It is not uncommon to need to derive one set of rasters from another. Raster math is pretty easy, but sometimes the steps required can be tricky. This trickiness includes computing a cumulative series, like cumulative SST.

Here we'll make a cumulative series of CHLOR_A values from our monthly 2018 dataset from the Gulf of Maine. There are a number of ways to do this, but we describe here a very simple approach. First we load packages from our library, then we load our database and filter it to just monthly CHLOR_A in 2018.

# Install ohwobpg package
installed <- rownames(installed.packages())
if (!("ohwobpg" %in% installed))  remotes::install_github("BigelowLab/ohwobpg", quiet = TRUE)
library(ohwobpg)
library(raster)
library(dplyr, warn.conflicts = FALSE)

path <- system.file("gom", package = "ohwobpg")
db <- read_database(path) %>%
  dplyr::filter(per == 'MO' &
                param == 'chlor_a')
db

So now we have our 12 records. Let's read them into a raster stack, chl.

chl <- raster::stack(as_filename(db, path))

Next we iterate through layers 2 through 12, computing the cumulative sum as we go. Afterwards, we'll give the layers more meaningful names.

for (i in seq(from = 2, to = raster::nlayers(chl))){
  chl[[i]] <- chl[[i]] + chl[[i-1]]
}
names(chl) <- format(db$date, "%b")
chl

Now we can do a simple plot for all 12 months. Note how missing data data propagates forward in time. We'll use the rasterVis package to draw, but if it isn't installed we'll just do a simple base-R plot that is similar.

installed <- rownames(installed.packages())
if ("rasterVis" %in% installed){
  rasterVis::levelplot(log10(chl))
} else {
  breaks <- seq(from = -1, to = 2.5, by = 0.5)
  # adapted from the [viridisLite]() package
  # pal <- viridisLite::magma(length(breaks) - 1)
  pal <- c("#000004FF", "#1D1147FF", "#51127CFF", "#822681FF", "#B63679FF", 
           "#E65164FF", "#FB8861FF", "#FEC287FF", "#FCFDBFFF")
  plot(log10(chl), col = pal, breaks = breaks)
}