R/event_timeseries_hourly.R

In LimnoTech/SWMPrStorm: Organizing, and Analyzing Estuary Monitoring Data for Storm Events

#' event_timeseries_hourly
#'
#' @param var_in .xlsx with all required input variables defined (string).
#' @param data_path pathway to cdmo data folder (string).
#' @param storm_nm name of storm event (string).
#' @param onset_start YYYY-MM-DD HH:MM:SS start of storm event onset period (string).
#' @param onset_end YYYY-MM-DD HH:MM:SS end of storm event onset period (string).
#' @param view_start YYYY-MM-DD HH:MM:SS first datetime of data to plot (string).
#' @param view_end YYYY-MM-DD HH:MM:SS last datetime of data to plot (string).
#' @param recovery_start YYYY-MM-DD HH:MM:SS start of storm recovery period (string).
#' @param recovery_end YYYY-MM-DD HH:MM:SS end of storm recovery period (string).
#' @param wq_sites comma separated list of wq stations to plot. if left blank, all active stations from reserve will be plotted. (string).
#' @param met_sites comma separated list of met stations to plot. if left blank, all active stations from reserve will be plotted. (string).
#' @param keep_flags comma separated list of data quality flags that should be kept (string).
#' @param reserve 3 digit reserve code (string).
#' @param skip TRUE/FALSE. If TRUE, function will be skipped (string).
#' @param user_units User defined units. Set to "English" or "SI". Default CDMO data is in SI units. Not all parameters have common equivalent English units (e.g. concentrations), and therefore not all will be converted.
#'
#'
#' @return plots are generated and saved in /output/wq/timeseries_event_hourly/ and /output/met/timeseries_event_hourly/
#' @export
#'
#' @examples
#'
#' \dontrun{
#' #StormVariables.xlsx is a template variable input file saved in data/
#' vars_in <- 'data/StormTrackVariables.xlsx'
#' single_storm_track(var_in = vars_in)
#' }
event_timeseries_hourly <- function(var_in,
                                    data_path = NULL,
                                    reserve = NULL,
                                    storm_nm = NULL,
                                    onset_start = NULL,
                                    onset_end = NULL,
                                    view_start = NULL,
                                    view_end = NULL,
                                    recovery_start = NULL,
                                    recovery_end = NULL,
                                    wq_sites = NULL,
                                    met_sites = NULL,
                                    keep_flags = NULL,
                                    skip = NULL,
                                    user_units = NULL){


  # ----------------------------------------------------------------------------
  # Define global variables
  # ----------------------------------------------------------------------------
  NERR.Site.ID_ <- rlang::sym('NERR.Site.ID')
  Status_ <- rlang::sym('Status')
  Station.Type_ <- rlang::sym('Station.Type')

  parameter_ <- rlang::sym('parameter')
  value_ <- rlang::sym('value')

  datetimestamp_ <- rlang::sym('datetimestamp')
  datetimestamp_day_ <- rlang::sym('datetimestamp_day')
  con_ <- rlang::sym('con')
  xmin_ <- rlang::sym('xmin')
  ymin_ <- rlang::sym('ymin')
  xmax_ <- rlang::sym('xmax')
  ymax_ <- rlang::sym('ymax')
  years_ <- rlang::sym('years')


  # ----------------------------------------------------------------------------
  # Read in Data
  # ----------------------------------------------------------------------------

  #a.  Read in the variable input template, var_in

  input_Parameters <- openxlsx::read.xlsx(var_in, sheet = "timeseries_hourly")
  input_Master <- openxlsx::read.xlsx(var_in, sheet = "MASTER")


  #b.  Read the following variables from template spreadsheet if not provided as optional arguments

  if(is.null(reserve)) reserve <- input_Master[1,2]


  stations <- get('sampling_stations') %>%
    dplyr::filter(!! NERR.Site.ID_ == reserve) %>%
    dplyr::filter(!! Status_ == "Active")

  wq_stations <- stations %>%
    dplyr::filter(!! Station.Type_ == 1)

  met_stations <- stations %>%
    dplyr::filter(!! Station.Type_ == 0)


  if(is.null(storm_nm)) storm_nm <- input_Parameters[1,2]
  if(is.null(onset_start)) onset_start <- input_Parameters[2,2]
  if(is.null(onset_end)) onset_end <- input_Parameters[3,2]
  if(is.null(view_start)) view_start <- input_Parameters[4,2]
  if(is.null(view_end)) view_end <- input_Parameters[5,2]
  if(is.null(recovery_start)) recovery_start <- input_Parameters[6,2]
  if(is.null(recovery_end)) recovery_end <- input_Parameters[7,2]
  #if(is.null(wq_sites)) wq_sites <- unlist(strsplit(input_Parameters[8,2],", "))
  if(is.null(wq_sites)) wq_sites <- if(is.na(input_Parameters[8,2])) {wq_stations$Station.Code} else {unlist(strsplit(input_Parameters[8,2],", "))}
  #if(is.null(met_sites)) met_sites <- unlist(strsplit(input_Parameters[9,2],", "))
  if(is.null(met_sites)) met_sites <- if(is.na(input_Parameters[9,2])) {met_stations$Station.Code} else {unlist(strsplit(input_Parameters[9,2],", "))}
  if(is.null(keep_flags)) keep_flags <- unlist(strsplit(input_Parameters[10,2],", "))
  if(is.null(skip)) skip <- input_Parameters[11,2]
  if(is.null(user_units)) user_units <- input_Parameters[12,2]
  if(is.null(data_path)) data_path <- 'data/cdmo'

  ############## Tests #########################################################
  if(skip == "TRUE") {return(warning("skip set to 'TRUE', skipping event_timeseries_hourly"))}


  # ----------------------------------------------------------------------------
  # MET
  # ----------------------------------------------------------------------------

  ### load, clean, and filter data

  data_type <- 'met'

  ls_par <- lapply(met_sites, SWMPr::import_local, path = data_path)
  ls_par <- lapply(ls_par, SWMPr::qaqc, qaqc_keep = keep_flags)
  ls_par <- lapply(ls_par, subset, subset = c(view_start, view_end))#, select = par) # Note: par <- wb_basic %>% .[[1]]


  names(ls_par) <- met_sites

  ## convert dataset to user defined units (if "SI", no conversion will take place)
  ls_par <- SWMPrStorm::convert_units(ls_par, user_units)

  ## add in precip intensity
  ls_par <- lapply(ls_par, function(x) {x$intensprcp <- x$totprcp * 4; x}) # unit/15-min to unit/hr


  ## identify parameters, remove a few
  parm <- unique(names(ls_par[[1]]))
  parm <- subset(parm, !(parm %in% c('datetimestamp', 'wdir', 'sdwdir', 'totpar', 'totsorad')))


  for(i in 1:length(ls_par)) {

    # select a station data frame & determine name
    dat <- ls_par[[i]]
    sta <- names(ls_par)[i]

    # tidy
    dat <- tidyr::pivot_longer(dat, cols = 2:12
                        , names_to = 'parameter'
                        , values_to = 'value')

    df_day <- dat %>%
      dplyr::filter(dplyr::between(!! datetimestamp_
                     , as.POSIXct(view_start)
                     , as.POSIXct(view_end))) %>%
      dplyr::mutate("datetimestamp_day" = lubridate::floor_date(!! datetimestamp_, unit = 'hour')) %>%
      dplyr::group_by(!! datetimestamp_day_, !! parameter_) %>%
      dplyr::summarize("value" = mean(!! value_, na.rm = T))

    for(j in 1:length(parm)) {

      # Create a dummy data.frame for events and recovery
      df <- data.frame(xmin=as.POSIXct(c(onset_start)),
                       xmax=as.POSIXct(c(onset_end)),
                       ymin=c(-Inf),
                       ymax=c(Inf),
                       years=c('Event Onset'))


      x <-
        df_day %>% dplyr::filter(!! parameter_ == parm[j])

      x <-
        ggplot2::ggplot(x, ggplot2::aes(x = !! datetimestamp_day_, y = !! value_)) +
        ggplot2::ggtitle(SWMPrExtension::title_labeler(nerr_site_id = sta)) +
        ggplot2::geom_line(ggplot2::aes(color = 'Hourly Avg'), lwd = 1) +# 'steelblue3') +
        ggplot2::geom_rect(data=df,ggplot2::aes(xmin=!! xmin_,ymin=!! ymin_,xmax=!! xmax_,ymax=!! ymax_,fill=!! years_),
                  alpha=0.1,inherit.aes=FALSE) +
        ggplot2::labs(x = '', y = SWMPrStorm::y_axis_unit_labeler(parm[j], user_units))

      x <-
        x +
        ggplot2::scale_color_manual('', values = c('steelblue3')) +
        ggplot2::scale_fill_manual('', values = c('steelblue3', 'green')) +
        ggplot2::scale_x_datetime(date_breaks = 'day', date_labels = '%b\n%d', guide = ggplot2::guide_axis(check.overlap = TRUE))


      x <- x +
        ggplot2::theme_bw() +
        ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5),
                       strip.background = ggplot2::element_blank(),
                       panel.grid = ggplot2::element_blank(),
                       panel.border = ggplot2::element_rect(color = 'black', fill = NA),
                       plot.margin = ggplot2::margin(5.5, 24, 5.5, 5.5, unit = 'pt'),
                       axis.title.y = ggplot2::element_text(margin = ggplot2::unit(c(0, 8, 0, 8), 'pt'), angle = 90),
                       text = ggplot2::element_text(size = 16),
                       legend.position = 'top')

      x_ttl <- paste('output/met/timeseries_event_hourly/timeseries_event_hourly_', sta, '_', parm[j], '.png', sep = '')

      ggplot2::ggsave(filename = x_ttl, plot = x, height = 4, width = 6, units = 'in', dpi = 300)

    }
  }



  # ----------------------------------------------------------------------------
  # WQ
  # ----------------------------------------------------------------------------


  ## load, clean, and filter data
  data_type <- 'wq'

  ls_par <- lapply(wq_sites, SWMPr::import_local, path = data_path)
  ls_par <- lapply(ls_par, SWMPr::qaqc, qaqc_keep = keep_flags)
  ls_par <- lapply(ls_par, subset, subset = c(view_start, view_end))

  names(ls_par) <- wq_sites

  ## convert dataset to user defined units (if "SI", no conversion will take place)
  ls_par <- SWMPrStorm::convert_units(ls_par, user_units)


  ## identify parameters
  parm <- unique(names(ls_par[[1]]))
  parm <- subset(parm, !(parm %in% c('datetimestamp')))

  # ----------------------------------------------
  # Time series, hourly smooth, event window -----
  # ----------------------------------------------

  for(i in 1:length(ls_par)) {

    # select a station data frame & determine name
    dat <- ls_par[[i]]
    sta <- names(ls_par)[i]

    # tidy
    dat <- tidyr::pivot_longer(dat, cols = 2:13
                        , names_to = 'parameter'
                        , values_to = 'value')

    df_day <- dat %>%
      dplyr::filter(dplyr::between(!! datetimestamp_
                     , as.POSIXct(view_start)
                     , as.POSIXct(view_end))) %>%
      dplyr::mutate("datetimestamp_day" = lubridate::floor_date(!! datetimestamp_, unit = 'hour')) %>%
      dplyr::group_by(!! datetimestamp_day_, !! parameter_) %>%
      dplyr::summarize("value" = mean(!! value_, na.rm = T))

    for(j in 1:length(parm)) {

      # Create a dummy data.frame for events and recovery
      df <- data.frame(xmin=as.POSIXct(c(onset_start)),
                       xmax=as.POSIXct(c(onset_end)),
                       ymin=c(-Inf),
                       ymax=c(Inf),
                       years=c('Event Onset'))

      x <-
        df_day %>%
        dplyr::filter(!! parameter_ == parm[j])

      x <-
        ggplot2::ggplot(x, ggplot2::aes(x = !! datetimestamp_day_, y = !! value_)) +
        ggplot2::ggtitle(SWMPrExtension::title_labeler(nerr_site_id = sta)) +
        ggplot2::geom_line(ggplot2::aes(color = 'Hourly Avg'), lwd = 1) +
        ggplot2::geom_rect(data=df,ggplot2::aes(xmin=!! xmin_,ymin=!! ymin_,xmax=!! xmax_,ymax=!! ymax_,fill=!! years_),
                  alpha=0.1,inherit.aes=FALSE) +
        ggplot2::labs(x = '', y = SWMPrStorm::y_axis_unit_labeler(parm[j],user_units)) #

      x <-
        x +
        ggplot2::scale_color_manual('', values = c('steelblue3')) +
        ggplot2::scale_fill_manual('', values = c('steelblue3', 'green')) +
        ggplot2::scale_x_datetime(date_breaks = 'day', date_labels = '%b\n%d', guide = ggplot2::guide_axis(check.overlap = TRUE))

      x <- x +
        ggplot2::theme_bw() +
        ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5),
                       strip.background = ggplot2::element_blank(),
                       panel.grid = ggplot2::element_blank(),
                       panel.border = ggplot2::element_rect(color = 'black', fill = NA),
                       plot.margin = ggplot2::margin(5.5, 24, 5.5, 5.5, unit = 'pt'),
                       axis.title.y = ggplot2::element_text(margin = ggplot2::unit(c(0, 8, 0, 8), 'pt'), angle = 90),
                       text = ggplot2::element_text(size = 16),
                       legend.position = 'top')

      x_ttl <- paste('output/wq/timeseries_event_hourly/timeseries_event_hourly_', sta, '_', parm[j], '_', storm_nm, '.png', sep = '')

      ggplot2::ggsave(filename = x_ttl, plot = x, height = 4, width = 6, units = 'in', dpi = 300)

    }
  }


}