Basic programming in R

EVR 628- Intro to Environmental Data Science

Author
Affiliation

Juan Carlos Villaseñor-Derbez (JC)

Rosenstiel School of Marine, Atmospheric, and Earth Science and Institute for Data Science and Computing

Part 0: Set up

  1. Go to the WCPFC website here.
  2. Towards the bottom of the page, look for a data set called Aggregated data, grouped by 1°x1° latitude/longitude grids, YEAR and MONTH.
  3. Click on green link (should say 3.33 MB) to download the data to your data/raw folder in your EVR 628 project
  4. You will have downloaded a ziped folder called WCPFC_S_PUBLIC_BY_1x1_MM_5.zip. Unzip it.
  5. Navigate to data/raw/WCPFC_S_PUBLIC_BY_1x1_MM_5 and open the pdf called purse_seine_by_mon_1x1.pdf.
  6. What does the metadata say about LAT_short and LON_short?

Part 1: Functions

Exercise 1: A function that cleans latitude / longitude data

For now, you will work with just one of the 12 data sets available in the WCPFC website. But if you had to work with all of them, you might want to build a function that deals with the

  1. Start a new R script called functions_and_loops.R, and save it to your scripts/processing folder
  2. Copy the code below to your new script (you can use the clipboard icon on the top right of the code chunk)
  3. Run everything to make sure it works. Take some time to inspect the wcpfc and the wcpfc_clean items.
  4. Look at the code. Does it look like some code is repeating? (Again. Imagine you will have to do this for other 11 data sets. And some of those will have a resolution of 5 vs 1 degree.)
# SET UP #######################################################################

## Load packages ---------------------------------------------------------------
library(tidyverse)
library(EVR628tools)

## Load data -------------------------------------------------------------------
wcpfc <- read_csv("data/raw/WCPFC_S_PUBLIC_BY_1x1_MM_5/WCPFC_S_PUBLIC_BY_1x1_MM.CSV")

# PROCESSING ###################################################################

## Clean data -------------------------------------------------------------------
wcpfc_clean <- wcpfc |> 
  select(year = yy,
         lat = lat_short,
         lon = lon_short,
         days) |> 
  # WCPFC reports "the latitude of the south-west corner"
  # We need to fix that
  mutate(
    # Find if it's in the S or W hemisphere
    lat_mult = ifelse(str_detect(lat, "N"), 1, -1),
    lon_mult = ifelse(str_detect(lon, "E"), 1, -1),
    # Find if it's in the S or W hemisphere
    lat = as.numeric(str_remove_all(lat, "[:alpha:]")),
    lon = as.numeric(str_remove_all(lon, "[:alpha:]")),
    # Find if it's in the S or W hemisphere
    lat = lat_mult * lat + 0.5,
    lon = lat_mult * lon + 0.5) |> 
  select(-lat_mult, -lon_mult)

wcpfc_clean
# A tibble: 193,063 × 4
    year   lat   lon  days
   <dbl> <dbl> <dbl> <dbl>
 1  1967   0.5  140.     0
 2  1967   0.5  142.     0
 3  1967  -0.5 -140.     0
 4  1967  -0.5 -142.     0
 5  1967   3.5  136.     0
 6  1967   7.5  134.     0
 7  1967   8.5  134.     0
 8  1968   0.5  140.     0
 9  1968   0.5  144.     0
10  1968  -0.5 -140.     0
# ℹ 193,053 more rows
  1. Let’s identify the general structure:
    1. Step 1: What are we repeating?
    2. Step 2: Within that code, what is constant and what changes?
  2. Let’s build a function called clean_coords()
Code 
clean_coords <- function(x, res) {
  mult <- ifelse(str_detect(x, "N|E"), 1, -1)
  coord <- as.numeric(str_remove_all(x, "[:alpha:]"))
  out <- (mult * coord) + (res / 2)
  return(out)
}
  1. Let’s test the function
Code 
# These are equivalent to 10, -10, 10, -10 for the bottom-left corner
test_coord <- c("10N", "10S", "10E", "10W")

# Once centered, they should be 10.5, -9.5, 10.5, -9.5
clean_coords(test_coord, res = 1)
[1] 10.5 -9.5 10.5 -9.5
  1. After confirming they perform, update the code to use our function
# SET UP #######################################################################

## Load packages ---------------------------------------------------------------
library(tidyverse)
library(EVR628tools)

## User defined functions ------------------------------------------------------
clean_coords <- function(x, res) {
  mult <- ifelse(str_detect(x, "N|E"), 1, -1)
  coord <- as.numeric(str_remove_all(x, "[:alpha:]"))
  out <- (mult * coord) + (res / 2)
  return(out)
}

## Load data -------------------------------------------------------------------
wcpfc <- read_csv("data/raw/WCPFC_S_PUBLIC_BY_1x1_MM_5/WCPFC_S_PUBLIC_BY_1x1_MM.CSV")

# PROCESSING ###################################################################

## Clean data -------------------------------------------------------------------
wcpfc_clean <- wcpfc |> 
  select(year = yy,
         lat = lat_short,
         lon = lon_short,
         days) |> 
  # WCPFC reports "the latitude of the south-west corner"
  # We need to fix that
  mutate(
    lat = clean_coords(lat, 1),
    lon = clean_coords(lon, 1))

wcpfc_clean
# A tibble: 193,063 × 4
    year   lat   lon  days
   <dbl> <dbl> <dbl> <dbl>
 1  1967   0.5  140.     0
 2  1967   0.5  142.     0
 3  1967  -0.5  142.     0
 4  1967  -0.5  142.     0
 5  1967   3.5  136.     0
 6  1967   7.5  134.     0
 7  1967   8.5  134.     0
 8  1968   0.5  140.     0
 9  1968   0.5  144.     0
10  1968  -0.5  142.     0
# ℹ 193,053 more rows

Exercise 2: A function that builds a plot

You are building 10 monographs for 10 different dive sites where lionfish are found. People are interested in knowing “how big are the lionfish in dive site X?”.

  1. Using the data_lionfish data, build a histogram of lionfish length (total_length_m) for all fish sampled from site == Paraiso. Don’t worry about themes and labels for now.
  2. Using the data_lionfish data, build a histogram of lionfish length (total_length_m) for all fish sampled from site == Canones. Don’t worry about themes and labels for now.
  3. Using the data_lionfish data, build a histogram of lionfish length (total_length_m) for all fish sampled from site == Paamul. Don’t worry about themes and labels for now.
  4. I think you see where I’m going. We can’t use facet_wrap or facet_grid because each dive site will have it’s own plot.
  5. Let’s identify the general structure:
    1. Step 1: What are we repeating?
    2. Step 2: Within that code, what is constant and what changes?
  6. Build a new function called site_histogram()
Code 
data("data_lionfish")

site_histogram <- function(data, my_site) {
  # Filter my data
  inside_data <- data |> 
    filter(site == my_site)
  
  # Build my plot
  p <- ggplot(inside_data,
              mapping = aes(x = total_length_mm)) +
    geom_histogram(binwidth = 50)
  
  # Return my plot
  return(p)
}

site_histogram(data = data_lionfish, 
               my_site = "Paraiso")

Part 2: for loops

We’ll do this one together.

  1. Build a for loop that:
    1. calls your site_histogram() function, and then
    2. saves it out as results/img/<site_name>.png
Code 
for(site_i in unique(data_lionfish$site)) {
  # Build plot with my function
  p <- site_histogram(data = data_lionfish,
                      my_site = site_i)
  
  # Save the ploit
  ggsave(plot = p,
         filename = paste("results/img/", site_i, ".png"),
         width = 6,
         height = 4)
  
  # End of item i
}