Deutscher Wetterdienst (DWD)

library(kwb.impetus)
dwd_timeperiod <- kwb.impetus::dwd_berlin_monthly %>%  
  dplyr::group_by(.data$parameter_name, 
               .data$parameter) %>%  
  dplyr::summarise(date_min = min(.data$date), 
                   date_max = max(.data$date), 
                   number_of_datapoints = dplyr::n()) %>%  
  dplyr::mutate(file_postfix = stringr::str_replace_all(.data$parameter, "_", "-"))

DT::datatable(dwd_timeperiod, caption = "Monthly DWD Data availability for Berlin")

summarise_means <- function(data) {
  data %>%
    dplyr::summarise(
      q05 = quantile(.data$mean, probs = 0.05),
      q95 = quantile(.data$mean, probs = 0.95),
      mean = mean(.data$mean)
    )  
}

filter_for_parameter <- function(data, parameter_name) {
  data[data$parameter_name == parameter_name,]
}

filter_for_max_year <- function(data) {
  data[data$Year == max(data$Year),]
}

aggregate_and_plot <- function(parameter_name, colors, unit = "mm/month") {
  
  full_data <- kwb.impetus::dwd_berlin_monthly %>% 
    filter_for_parameter(parameter_name) %>%  
    dplyr::mutate(
      Year = as.integer(.data$year), 
      Decade = kwb.impetus::floor_decade(.data$year),
      Decade_Label = kwb.impetus::decade_label(.data$Decade),
      Month = as.factor(.data$month),
      Label = sprintf("DWD, monthly '%s'", parameter_name)
    )
  
  aggregated_data <- full_data %>% 
    kwb.impetus::group_by_decade_month_label() %>% 
    summarise_means()
  
  decades <- kwb.impetus::decades_tibble(
    decade_labels = aggregated_data$Decade_Label,
    colors = colors
  )
  
  decade_mean_data <- aggregated_data %>%  
    kwb.impetus::group_by_decade_label() %>% 
    dplyr::summarise(annual_mean = sum(.data$mean))
  
  # p1 <- aggregated_data %>%
  #   dplyr::left_join(decade_mean_data) %>%
  #   dplyr::mutate(Decade_Label = sprintf(
  #     "%s\n%3.1f mm/Jahr)",
  #     .data$Decade_Label,
  #     round(.data$annual_mean, 1)
  #   ))
  
  p1 <- aggregated_data
  
  p2 <- full_data %>% 
    filter_for_max_year() %>% 
    kwb.impetus::group_by_decade_label() %>% 
    kwb.impetus::group_by_year_month() %>% 
    summarise_means()
  
  p1 %>% 
    ggplot2::ggplot(mapping = ggplot2::aes(
      x = as.integer(.data$Month), 
      y = .data$mean,
      col = as.factor(.data$Decade_Label)
    )) + 
    #ggplot2::geom_point(alpha = 0.5) + 
    kwb.impetus::decade_ribbons() + 
    kwb.impetus::scale_fill_decades(decades) +
    kwb.impetus::scale_color_decades(decades) +
    ggplot2::geom_point() +
    ggplot2::geom_point(ggplot2::aes(
      x = as.integer(.data$Month),
      y = .data$mean
    ), 
    data = p2,
    col = "darkgrey",
    alpha = 0.5, 
    inherit.aes = FALSE,
    show.legend = FALSE
    ) +
    kwb.impetus::ggplot2_scale_x_continuous_12() +
    #ggplot2::geom_boxplot() +
    ggplot2::facet_wrap( 
      ~ .data$Decade_Label,
      nrow = 1L, 
      ncol = length(unique(p1$Decade_Label))
    ) +
    ggplot2::theme_bw() + 
    ggplot2::theme(legend.position = "bottom") +
    ggplot2::labs(
      y = sprintf("Mean '%s' (%s)", 
                  parameter_name, 
                  unit), 
      x = "Month Number", 
      col = "Mean", 
      fill = "5%/95% Conf.-Interval",
      title = unique(p1$Label)
    )
}

print_to_pdf <- function(gg, file, width.cm) {
  
  kwb.utils::preparePdf(
    file, 
    landscape = TRUE, 
    width.cm = width.cm, 
    height.cm = 21
  )

  on.exit(dev.off())
  
  print(gg)
}