Disney_Marathon_Weather.R

library(tidyverse)
library(reshape2)
library(lubridate)
library(rnoaa)

token <- '' # Get token here: https://www.ncdc.noaa.gov//cdo-web/token
options(noaakey = token)

marathon_dates <- c('2021-01-10','2020-01-12','2019-01-13',
					'2018-01-07','2017-01-08','2016-01-10','2015-01-11','2014-01-12',
					'2013-01-13','2012-01-08','2011-01-09','2010-01-10','2009-01-11',
					'2008-01-13','2007-01-07','2006-01-08','2005-01-09','2004-01-11',
					'2003-01-12','2002-01-06','2001-01-07','2000-01-09','1999-01-10',
					'1998-01-11','1997-01-05','1996-01-07','1995-01-08','1994-01-16') %>%
	as.Date()


# Find the closest weather station
x <- isd_stations_search(lat = 28.3772, lon = -81.5707, radius = 17)
x

weather <- data.frame(date = marathon_dates,
					  min_temp = NA_real_,
					  max_temp = NA_real_,
					  start_temp = NA_real_,
					  avg_temp = NA_real_,
					  avg_wind_speed = NA_real_) %>%
	mutate(year = year(date))
for(i in 1:nrow(weather)) {
	weather_data <- isd(usaf = x[1,]$usaf, wban = x[1,]$wban, year = year(weather[i,]$date)) %>%
		dplyr::filter(temperature != '+9999') %>% ## remove +9999's
		dplyr::mutate(temp_celsuis = as.numeric(temperature) / 10,
					  temp_fahrenheit = temp_celsuis * 9/5 +32,
					  date_time = parse_date_time(paste0(date, time), 'y-m-d HM', tz = 'America/New_York'),
					  date = as.Date(date, format = '%Y%m%d'),
					  wind_speed = 1.60934 * as.numeric(wind_speed) / 10,
					  time_from_start = abs(date_time - as.POSIXct(paste0(date, ' 05:30:00'), tz = 'America/New_York'))
					) %>%
		dplyr::filter(date == weather[i,]$date) %>%
		select(date, time, date_time, temp_celsuis, temp_fahrenheit, wind_speed, time_from_start)
	
	if(nrow(weather_data) > 0) {
		weather[i,]$min_temp <- min(weather_data$temp_fahrenheit, na.rm = TRUE)
		weather[i,]$max_temp <- max(weather_data$temp_fahrenheit, na.rm = TRUE)
		weather[i,]$start_temp <- weather_data[weather_data$time_from_start == min(weather_data$time_from_start),]$temp_fahrenheit
		weather[i,]$avg_temp <- mean(weather_data$temp_fahrenheit, na.rm = TRUE)
		weather[i,]$avg_wind_speed <- mean(weather_data$wind_speed, na.rm = TRUE)
	} else {
		print(paste0('No data found for ', weather[i,]$date))
	}
}

weather <- weather[complete.cases(weather),]
weather.melt <- melt(weather[,c('year', 'min_temp', 'max_temp', 'start_temp', 'avg_temp', 'avg_wind_speed')],
					 id.vars = c('year', 'start_temp', 'avg_temp', 'avg_wind_speed'))
ggplot(weather.melt, aes(x = year)) + 
	geom_ribbon(data = weather, aes(ymin = min_temp, ymax = max_temp), alpha = 0.3, fill = 'skyblue') +
	geom_path(aes(y = start_temp, group = variable), color = 'maroon') +
	geom_point(aes(y = start_temp), color = 'maroon') +
	geom_path(aes(y = value, group = variable)) +
	geom_point(aes(y = value, size = avg_wind_speed)) +
	geom_text(data = weather, aes(y = max_temp, label = paste0(max_temp, '°')), vjust = -1) +
	geom_text(data = weather, aes(y = min_temp, label = paste0(min_temp, '°')), vjust = 2) +
	scale_x_continuous(breaks = seq(min(weather$year), max(weather$year))) +
	scale_size_continuous('Wind\nSpeed\n(mph)') +
	ylim(c(25, 85)) +
	ylab('Temperature (Fahrenheit)') + xlab('') + theme_minimal() +
	ggtitle('Disney Marathon Weather',
			subtitle = 'Temperature range and starting temperatures (in red)')

ggsave('DisneyMarathonWeather.png', width = 10, height = 4)