Anomaly Example

Load libraries.

library(stringr)
library(dplyr)
library(azmpdata)
library(multivaR)
#source('anomaly.R')
library(ggplot2)
library(gridExtra, warn.conflicts = FALSE)

Get test data files from Benoit. Once azmpdata is finished this section will be replaced by a query to azmpdata.

# test files
 fp <- 'C:/Users/chisholme/desktop/April2020- March2021/Data Access/draft_csv/'
 list_fn <- list.files(fp)

 # list_fn <- 'Discrete_Occupations_Sections.csv'


# load in csv

for (i in 1:length(list_fn)){
  # isolate df name

  dfname <- gsub('(\\w+)\\.csv', '\\1', list_fn[i]) # thanks CL for the reg expression!
    #dfname <- 'Derived_Occupations_Sections'
  eval(parse(text = paste(dfname, '<- read.csv(file.path(fp, list_fn[i]))')))

}

head(Derived_Occupations_Sections)

NA

Calculate monthly and annual anomalies.Modified versions of Chantelle’s functions.

# test anomaly calcs

dd <- monthlyAnomaly(d = Derived_Occupations_Sections, climatologyYears = c(1999, 2010), var = 'integrated_chlorophyll_0_100')

dd <- monthlyAnomaly(d = dd, climatologyYears = c(1999, 2010), var = 'integrated_nitrate_0_50')

# test annual results

ddann <- annualAnomaly(d = dd, anomaly = 'integrated_chlorophyll_0_100_anomaly')

head(ddann)

Improvements to anomaly functions:

• account for possibility of data collection stretching between two months, and wanting one anomaly value (from one cruise)
  • once cruise_id is incorporated into metadata an option will be written into the function to group data by cruise rather than month
• Ability to split data by section or station within data frame
  • Benoit’s scripts split data by section or station before calculating anomalies
• Difference in calculations between Benoit and Chantelle versions
  • Due to difference in calculating monthly vs annual climatology

# Benoit's method

df_data_filtered_l <- Derived_Occupations_Sections

df_data_filtered_l <- df_data_filtered_l %>%
  select(., section, station, latitude, longitude, year, month, day, event_id, integrated_chlorophyll_0_100)

##--------------------------------------------------------------------------------------------
## annual climatology
df_climatology_annual_l <- df_data_filtered_l %>%
  dplyr::filter(., year>=1999, year<=2010) %>%
  dplyr::group_by(.,  year) %>% # removed group by section
  dplyr::summarise(., mean=mean(integrated_chlorophyll_0_100, na.rm=TRUE), sd=sd(integrated_chlorophyll_0_100, na.rm=TRUE)) %>%
  dplyr::ungroup(.)

`summarise()` ungrouping output (override with `.groups` argument)

  ##--------------------------------------------------------------------------------------------
## annual anomalies
df_anomaly_annual_l <- dplyr::left_join(df_data_filtered_l,
                          df_climatology_annual_l ,
                          by=c( "year")) %>% # removed section
  dplyr::group_by(., year) %>%
  dplyr::mutate(., value=(integrated_chlorophyll_0_100-mean)/sd) %>%
  summarise(., annual_anomaly = mean(value, na.rm = TRUE)) %>%
  dplyr::select(., year,  annual_anomaly) %>%
  dplyr::ungroup(.)

`summarise()` ungrouping output (override with `.groups` argument)

# compare Benoit and Chantelle methods

anncom <- cbind(df_anomaly_annual_l, ddann)
par(mar = c(5, 4, 4, 4) + 0.3)  # Leave space for z axis
plot(anncom$year, anncom$integrated_chlorophyll_0_100_anomaly, type = 'l', col = 'red', ylab = '', xlab = 'Year') # first plot
par(new = TRUE)

axis(side = 2, col = 'red')
par(new = TRUE)

plot(anncom$year, anncom$annual_anomaly, type = "l", col = 'blue', axes = FALSE, bty = "n", xlab = "", ylab = "")
axis(side=4, at = pretty(range(anncom$annual_anomaly, na.rm = TRUE)), col = 'blue')

mtext("annual anomaly - Benoit", side=4, line=3, col = 'blue')
mtext("annual anomaly - Chantelle", side=2, line=3, col = 'red')

NA
NA