KNMI NL weerdata graphing

weatherdata.R

# visualise weather in NL
#
#
# (c) Jules Stuifbergen 2018 - Creative Commons licence CC BY-SA
#
#

library(httr)
library(tidyverse)
library(lubridate)
library(viridis)
library(gganimate)
library(animation)


datafile <- "./weerdata.txt"

# download raw data
knmi_url <- 'http://projects.knmi.nl/klimatologie/daggegevens/getdata_dag.cgi'
knmi_params <- list(start = "19000101",
                    end   = format(Sys.Date(), "%Y%m%d"),
                    vars  = "TEMP",
                    stns  = "260")        #station 260 = De Bilt

knmi_req <- GET(knmi_url, query = knmi_params) # documentation says POST, but that doesn't work

# write datafile 
write_file(content(knmi_req, as = "text"), datafile)

# output:
# STN,YYYYMMDD,   TG,   TN,  TNH,   TX,  TXH, T10N,T10NH

# read datafile
knmi_raw <- read_csv(datafile, comment = "#",
                     col_names = c("station", "yyyymmdd", "mean", "min", "min_h", "max", "max_h", "t10_min", "t10_min_h"))

# cleanup raw datafile, select some vars, make date, convert to actual degrees
knmi <- knmi_raw %>% transmute( dag = as_date(as.character(yyyymmdd)), jaar = as.character(year(dag)),
                                doy = yday(dag),
                                gem = mean/10, min = min/10, min_h = min_h,
                                max = max/10, max_h = max_h,
                                t10_min = as.numeric(t10_min)/10, t10_min_h = t10_min_h,
                                soort_dag = case_when( max >= 30 ~ "tropische dag",
                                                       max >= 25 ~ "hete dag",
                                                       max >= 20 ~ "warme dag")
                              )

# plot some cool stuff

# animated barplot
p1 <- ggplot( knmi %>% mutate( max_temp = round(max)) %>%                   # vanwege de kleur moet dit even met de hand geteld worden
               group_by(jaar, max_temp) %>% summarise(n = n()),
             aes( x = max_temp, y = n, frame = jaar) ) +
  geom_bar(aes(fill = max_temp), stat = "identity", position = "identity", width = 1) +
  scale_fill_viridis(discrete = FALSE, option = "B") +
  theme_minimal() + theme(legend.position = "none") +
  xlab("maximum dagtemperatuur") +
  ylab("aantal dagen dit jaar met deze temperatuur")

ani.options(interval = 0.25)
gganimate(p1)


# hot days - stacked barplot of freq

# dataset as is, count days (barplot)
ggplot( filter(knmi, !is.na(soort_dag)), aes( x = factor(jaar))) +
  geom_bar(aes(fill = factor(soort_dag, levels = c("warme dag","hete dag","tropische dag"))), width=1) +
  scale_fill_manual(values = c("tropische dag" = "red", "hete dag" = "darkorange", "warme dag" = "gold"), name = "soort dag") +
  theme_minimal() + theme(axis.text.x=element_text(angle=60, hjust=1)) +
  scale_x_discrete(breaks = seq(1900, 2020, 5)) + 
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank()) +
  xlab("")  + ylab("aantal van deze dagen in het gegeven jaar") +
  ggtitle("Tropische, hete en warme dagen sinds 1900")


# line graph follows.. first count manually
linedata <- knmi %>%
  filter(!is.na(soort_dag), jaar < max(jaar)) %>%  # remove this (incomplete) year
  group_by(jaar, soort_dag) %>%
  summarise(aantal = n())


ggplot(linedata, aes(x = jaar, y = aantal)) +
  geom_line(aes(color = soort_dag, group = soort_dag)) +
  geom_smooth(aes(color = soort_dag, group = soort_dag, alpha = .2), fill = "lightgrey", alpha = .6, linetype = "dotted") +
  scale_color_manual(values = c("tropische dag" = "red", "hete dag" = "darkorange", "warme dag" = "gold"), name = "soort dag") +
  theme_minimal() + theme(axis.text.x=element_text(angle=60, hjust=1)) +
  scale_x_discrete(breaks = seq(1900, 2020, 5)) + 
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank()) +
  xlab("")  + ylab("aantal van deze dagen in het gegeven jaar") +
  ggtitle("Tropische, hete en warme dagen sinds 1900")