Tidy-Tuesday Geospatial

Ibrahim Uruc Tarim

---

first_custom <- function() {
  theme_minimal() +
    theme(
      axis.line = element_line(color = "gray0"),
      plot.background  = element_rect(fill = "grey95"),
      panel.grid.major = element_line(linewidth = 0.5, color = "gray80"),
      panel.grid.minor = element_blank(),
      #panel.grid.minor = element_line(color = "gray85"),
      plot.margin = margin(15, 15, 15 ,15),
      plot.title = element_text(face = "bold"),
      plot.title.position = "plot",
      plot.caption = element_text(size = 8, face = "italic"),
      plot.caption.position = "plot",
      legend.background = element_rect(),
      legend.justification = (c(1,0)),
      legend.text = element_text(size = 8),
      legend.title = element_text(size = 8),
      legend.key.size = unit(0.25, "cm")
    )
}

Tidy Tuesday Description

Taken from the GitHub repo

The TidyTuesday Project is a weekly data project aimed at the R ecosystem. As this project was born out of the R4DS Online Learning Community and the R for Data Science textbook, an emphasis was placed on understanding how to summarize and arrange data to make meaningful charts with ggplot2, tidyr, dplyr, and other tools in the tidyverse ecosystem, but other packages are more than welcome.

Every week they post a raw dataset, a chart or article related to that dataset, and ask participants to explore the data. While the dataset will be “tamed”, it will not always be tidy! As such you might need to apply various R for Data Science techniques to wrangle the data into a true tidy format. For this lab, we won’t require you to do much cleaning. The goal of TidyTuesday is to apply your R skills, get feedback, explore other’s work, and connect with the greater #RStats community!

TidyTuesday is intended to be a safe and supportive environment.

---

Choosing Data, Exploring/Cleaning, and Tables

The TidyTuesday Repo contains weekly datasets for the last four years. There is a lot of documentation on the repo, please check it out. If you’re looking for some inspiration, #TidyTuesday is a popular hashtag on Twitter and other platforms.

• Some datasets have a clean version, please use this one if its available.
• Some TidyTuesday folders have multiple datasets in them, you can use one or all of these. Just make sure your analysis includes the required components.

Lets Find Some Data

rainfall <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2020/2020-01-07/rainfall.csv')
temperature <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2020/2020-01-07/temperature.csv')

temperature <- temperature %>% 
  mutate(year = as.numeric(format(date, "%Y",)),
         month = as.numeric(format(date, "%m")))

temperature$city_name = toupper(temperature$city_name)
rainfall$city_name = toupper(rainfall$city_name)

Rainfall and Temperature data of Australia https://github.com/rfordatascience/tidytuesday/blob/master/data/2020/2020-01-07/readme.md

library(viridis)

# Find the intersection of city names
common_cities <- intersect(temperature$city_name, rainfall$city_name)

# Filter both datasets to keep only the common cities
temperature <- temperature %>%
  filter(city_name %in% common_cities)
rainfall <- rainfall %>%
  filter(city_name %in% common_cities)

temperature <- temperature %>% 
  mutate(year = as.numeric(format(date, "%Y",)),
         month = as.numeric(format(date, "%m")))

temperature$city_name = toupper(temperature$city_name)
rainfall$city_name = toupper(rainfall$city_name)

temp_recent_mean <- temperature %>%
  filter(year >= 2000, year <= 2019) %>%
  mutate(city_name = str_to_title(city_name)) %>%
  filter(temp_type == "max") %>%
  group_by(year, city_name) %>%
  summarise(temperature = mean(temperature, na.rm = TRUE), .groups='drop')

ggplot(temp_recent_mean, aes(fill=city_name, x=year, y=temperature, color=city_name)) +
  geom_line() +
  facet_wrap(~city_name, nrow=3) +
  scale_x_continuous(name="Year", breaks=seq(2000, 2019, 5)) +
  scale_y_continuous(name="Max Temp in Celsius", breaks=seq(14, 30, 2)) +
  ggtitle("Temperature Variation in Each City: 2000-2019") +
  theme(legend.position="none")

rain_recent_mean <- rainfall %>%
  filter(year >= 2000, year <= 2019) %>%
  group_by(year, city_name) %>%
  summarise(mean_rain = mean(rainfall, na.rm=TRUE), .groups='drop') %>%
  arrange(desc(mean_rain))

ggplot(rain_recent_mean, aes(fill=city_name, x=year, y=mean_rain, color=city_name)) +
  geom_line() +
  geom_point() +
  scale_x_continuous(name="Year", breaks=seq(2000, 2019, 5)) +
  scale_y_continuous(name="Average Rain in mm", breaks=seq(1, 8, 1)) +
  facet_wrap(~city_name) +
  ggtitle("Average Rainfall in Cities: 2000-2019") +
  theme(legend.position="none", axis.text.x=element_text(size=6))

## Tables

library(dplyr)
library(knitr)

# Find common cities
common_cities <- intersect(temperature$city_name, rainfall$city_name)

# Filter datasets to keep only common cities and specified years
temp_recent_common <- temperature %>%
  filter(city_name %in% common_cities, year >= 2000, year <= 2019)

rain_recent_common <- rainfall %>%
  filter(city_name %in% common_cities, year >= 2000, year <= 2019)

# Calculate temperature summary statistics
summary_stats_temp <- temp_recent_common %>%
  group_by(city_name, year) %>%
  summarize(
    mean_temp = mean(temperature, na.rm = TRUE),
    median_temp = median(temperature, na.rm = TRUE),
    min_temp = min(temperature, na.rm = TRUE),
    max_temp = max(temperature, na.rm = TRUE),
    .groups = 'drop'
  )

# Calculate rainfall summary statistics
summary_stats_rain <- rain_recent_common %>%
  group_by(city_name, year) %>%
  summarize(
    mean_rain = mean(rainfall, na.rm = TRUE),
    median_rain = median(rainfall, na.rm = TRUE),
    percentile_75_rain = quantile(rainfall, 0.75, na.rm = TRUE),
    total_rain = sum(rainfall, na.rm = TRUE),
    .groups = 'drop'
  )

# Output summary tables
kable(summary_stats_temp, "html") %>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed", "responsive")) %>%
  scroll_box(width = "100%", height = "500px")

city_name	year	mean_temp	median_temp	min_temp	max_temp
BRISBANE	2000	20.08620	20.80	2.5	37.4
BRISBANE	2001	20.45110	21.60	2.6	36.7
BRISBANE	2002	20.46534	21.10	1.6	36.0
BRISBANE	2003	20.17521	21.00	1.8	35.1
BRISBANE	2004	20.63497	21.85	2.5	40.2
BRISBANE	2005	20.86644	21.60	4.1	35.4
BRISBANE	2006	20.46356	21.70	3.0	32.2
BRISBANE	2007	20.63767	21.40	-0.1	34.2
BRISBANE	2008	20.07910	20.90	2.6	39.8
BRISBANE	2009	20.90932	21.70	3.4	34.8
BRISBANE	2010	20.53000	21.15	5.0	32.1
BRISBANE	2011	20.06027	21.15	3.2	33.5
BRISBANE	2012	20.23844	21.00	3.7	35.9
BRISBANE	2013	20.69411	21.35	4.1	32.4
BRISBANE	2014	20.95918	21.80	2.6	34.3
BRISBANE	2015	20.65151	21.60	4.5	36.3
BRISBANE	2016	21.18470	21.65	4.6	36.3
BRISBANE	2017	21.23452	22.30	4.3	34.9
BRISBANE	2018	20.77205	21.70	3.3	38.1
BRISBANE	2019	23.43079	23.50	5.2	34.0
CANBERRA	2000	13.05683	13.10	-7.9	35.1
CANBERRA	2001	13.33278	13.90	-6.8	39.9
CANBERRA	2002	13.56096	13.35	-7.0	36.9
CANBERRA	2003	13.52452	13.40	-5.7	39.8
CANBERRA	2004	13.76325	13.20	-8.3	38.7
CANBERRA	2005	13.88137	13.55	-5.4	39.4
CANBERRA	2006	13.97466	14.05	-7.4	39.9
CANBERRA	2007	14.31397	13.60	-6.1	40.5
CANBERRA	2008	13.24727	13.00	-6.7	35.8
CANBERRA	2009	14.18411	13.20	-7.5	40.0
CANBERRA	2010	13.54262	13.70	-6.5	39.1
CANBERRA	2011	13.30836	14.00	-8.0	37.5
CANBERRA	2012	12.84221	13.20	-6.8	35.1
CANBERRA	2013	13.97411	13.70	-6.0	41.6
CANBERRA	2014	14.08233	13.70	-7.6	40.2
CANBERRA	2015	13.74609	13.50	-7.0	37.0
CANBERRA	2016	14.39098	14.00	-5.5	39.3
CANBERRA	2017	14.02428	13.80	-8.7	41.6
CANBERRA	2018	14.43000	13.70	-7.4	40.6
CANBERRA	2019	18.18775	17.40	-2.7	41.6
MELBOURNE	2000	15.94877	15.10	1.8	38.9
MELBOURNE	2001	15.65863	14.75	3.6	40.9
MELBOURNE	2002	15.66014	14.90	1.2	36.7
MELBOURNE	2003	15.54137	14.90	1.4	43.9
MELBOURNE	2004	15.34071	14.40	1.4	39.7
MELBOURNE	2005	16.05630	15.20	1.9	42.7
MELBOURNE	2006	15.43247	14.70	2.2	42.2
MELBOURNE	2007	16.53178	15.50	2.0	40.9
MELBOURNE	2008	15.65451	14.70	2.3	41.0
MELBOURNE	2009	16.19822	15.30	1.8	45.9
MELBOURNE	2010	15.93580	15.10	3.0	43.4
MELBOURNE	2011	15.82593	15.20	2.5	39.5
MELBOURNE	2012	15.79302	14.80	2.1	39.7
MELBOURNE	2013	16.27411	15.60	1.3	40.9
MELBOURNE	2014	16.47233	15.45	1.2	43.4
MELBOURNE	2015	15.91562	14.70	0.6	41.2
MELBOURNE	2016	16.17637	15.50	3.0	42.2
MELBOURNE	2017	16.26849	15.40	0.8	37.8
MELBOURNE	2018	16.22466	15.30	0.8	41.7
MELBOURNE	2019	18.98510	18.05	6.2	42.8
PERTH	2000	18.75533	18.50	0.4	41.1
PERTH	2001	18.07329	18.00	0.5	40.1
PERTH	2002	18.62493	18.40	1.0	43.2
PERTH	2003	18.81836	18.20	1.2	43.8
PERTH	2004	18.59385	18.20	0.1	42.1
PERTH	2005	17.92027	17.65	0.8	42.6
PERTH	2006	18.58904	18.50	-1.3	40.9
PERTH	2007	18.41027	18.20	2.5	44.5
PERTH	2008	18.57542	18.50	1.3	43.4
PERTH	2009	18.89589	17.80	0.9	43.0
PERTH	2010	18.91205	18.50	-1.0	43.2
PERTH	2011	19.79918	19.25	0.2	41.8
PERTH	2012	19.14686	18.85	-0.7	42.2
PERTH	2013	19.34315	18.95	-0.3	41.1
PERTH	2014	19.21041	18.70	1.0	43.5
PERTH	2015	19.43534	18.55	0.1	44.2
PERTH	2016	18.20505	17.70	0.9	43.0
PERTH	2017	18.95603	18.10	1.0	43.3
PERTH	2018	18.77174	18.20	2.5	40.9
PERTH	2019	21.34636	21.15	1.5	42.9
SYDNEY	2000	18.66585	19.00	5.6	38.6
SYDNEY	2001	18.92712	19.15	5.8	35.8
SYDNEY	2002	18.85995	18.90	4.3	38.5
SYDNEY	2003	18.61479	19.10	4.9	39.6
SYDNEY	2004	19.08142	19.40	5.9	39.9
SYDNEY	2005	19.09877	19.40	5.1	39.2
SYDNEY	2006	19.05603	19.30	5.8	45.0
SYDNEY	2007	19.09699	19.40	3.7	39.5
SYDNEY	2008	18.34891	18.70	5.3	35.5
SYDNEY	2009	19.18459	19.60	5.9	41.7
SYDNEY	2010	18.99712	19.00	4.3	42.1
SYDNEY	2011	18.66575	18.80	5.5	41.5
SYDNEY	2012	18.51421	18.90	5.5	34.2
SYDNEY	2013	19.37041	19.90	7.0	45.8
SYDNEY	2014	19.27912	19.55	5.5	36.5
SYDNEY	2015	19.03712	19.30	5.0	40.9
SYDNEY	2016	19.63251	19.85	5.4	39.2
SYDNEY	2017	19.44507	19.85	5.4	39.4
SYDNEY	2018	19.14033	19.50	4.6	43.4
SYDNEY	2019	22.03046	22.15	7.8	39.6

kable(summary_stats_rain, "html") %>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed", "responsive")) %>%
  scroll_box(width = "100%", height = "500px")

city_name	year	mean_rain	median_rain	percentile_75_rain	total_rain
BRISBANE	2000	1.9666667	0.0	0.600	743.4
BRISBANE	2001	2.9766162	0.0	0.400	2164.0
BRISBANE	2002	1.8958621	0.0	0.000	1374.5
BRISBANE	2003	2.3370474	0.0	0.200	1678.0
BRISBANE	2004	3.2303797	0.0	0.000	2296.8
BRISBANE	2005	2.0973050	0.0	0.200	1478.6
BRISBANE	2006	2.1753463	0.0	0.200	1570.6
BRISBANE	2007	1.9217391	0.0	0.400	1370.2
BRISBANE	2008	5.0038544	0.0	4.000	2336.8
BRISBANE	2009	4.6768018	0.0	2.200	2076.5
BRISBANE	2010	7.4293850	0.2	4.800	3261.5
BRISBANE	2011	4.7905312	0.0	3.600	2074.3
BRISBANE	2012	5.0582927	0.0	2.350	2073.9
BRISBANE	2013	4.8500000	0.0	2.200	2027.3
BRISBANE	2014	2.4940129	0.0	0.200	1541.3
BRISBANE	2015	5.1922727	0.0	0.400	2284.6
BRISBANE	2016	2.0925620	0.0	0.400	759.6
BRISBANE	2017	2.7191011	0.0	0.200	968.0
BRISBANE	2018	2.4689655	0.0	0.850	859.2
BRISBANE	2019	1.8033058	0.0	0.200	654.6
CANBERRA	2008	2.2940000	0.0	0.050	229.4
CANBERRA	2009	1.1819178	0.0	0.200	431.4
CANBERRA	2010	2.6290411	0.0	0.400	959.6
CANBERRA	2011	1.5870879	0.0	0.400	577.7
CANBERRA	2012	1.8950820	0.0	0.200	693.6
CANBERRA	2013	1.4624658	0.0	0.200	533.8
CANBERRA	2014	1.5663912	0.0	0.200	568.6
CANBERRA	2015	1.5113573	0.0	0.200	545.6
CANBERRA	2016	2.1535519	0.0	0.800	788.2
CANBERRA	2017	1.3388430	0.0	0.200	486.0
CANBERRA	2018	1.2967033	0.0	0.000	472.0
CANBERRA	2019	0.9824658	0.0	0.000	358.6
MELBOURNE	2000	1.6349398	0.0	0.600	542.8
MELBOURNE	2001	1.8050147	0.0	1.200	611.9
MELBOURNE	2002	1.2055249	0.0	0.775	436.4
MELBOURNE	2003	1.3473973	0.0	0.800	491.8
MELBOURNE	2004	1.6473829	0.0	0.900	598.0
MELBOURNE	2005	1.5087912	0.0	0.800	549.2
MELBOURNE	2006	1.0264264	0.0	0.400	341.8
MELBOURNE	2007	1.2233533	0.0	0.800	408.6
MELBOURNE	2008	1.1803279	0.0	0.600	432.0
MELBOURNE	2009	1.1369863	0.0	0.600	415.0
MELBOURNE	2010	1.9949580	0.0	1.000	712.2
MELBOURNE	2011	2.3900277	0.0	1.000	862.8
MELBOURNE	2012	1.6983516	0.0	1.600	618.2
MELBOURNE	2013	1.8327824	0.0	1.400	665.3
MELBOURNE	2014	1.3164835	0.0	0.800	479.2
MELBOURNE	2015	1.2922636	0.0	0.600	451.0
MELBOURNE	2016	1.9286932	0.0	1.450	678.9
MELBOURNE	2017	1.7706371	0.0	1.000	639.2
MELBOURNE	2018	1.4876033	0.0	0.600	540.0
MELBOURNE	2019	1.2196133	0.0	1.200	441.5
PERTH	2000	1.9900826	0.0	0.650	722.4
PERTH	2001	1.9284507	0.0	0.000	684.6
PERTH	2002	1.7083102	0.0	0.200	616.7
PERTH	2003	2.1250689	0.0	0.800	771.4
PERTH	2004	1.5684211	0.0	0.400	566.2
PERTH	2005	2.2229917	0.0	0.900	802.5
PERTH	2006	1.2200000	0.0	0.000	445.3
PERTH	2007	1.8549451	0.0	0.825	675.2
PERTH	2008	2.1204420	0.0	0.875	767.6
PERTH	2009	1.5241758	0.0	0.400	554.8
PERTH	2010	1.4082418	0.0	0.000	512.6
PERTH	2011	2.1866667	0.0	0.200	787.2
PERTH	2012	1.7879656	0.0	0.200	624.0
PERTH	2013	2.0713889	0.0	0.425	745.7
PERTH	2014	1.7834254	0.0	0.000	645.6
PERTH	2015	1.5380556	0.0	0.000	553.7
PERTH	2016	2.0519663	0.0	0.400	730.5
PERTH	2017	2.0490028	0.0	0.200	719.2
PERTH	2018	2.1263736	0.0	0.200	774.0
PERTH	2019	1.5082192	0.0	0.000	550.5
SYDNEY	2000	2.2448087	0.0	0.800	821.6
SYDNEY	2001	3.7227397	0.0	1.200	1358.8
SYDNEY	2002	2.3561644	0.0	0.400	860.0
SYDNEY	2003	3.2887671	0.0	1.200	1200.4
SYDNEY	2004	2.7191257	0.0	0.550	995.2
SYDNEY	2005	2.2356164	0.0	0.600	816.0
SYDNEY	2006	2.7232877	0.0	1.000	994.0
SYDNEY	2007	4.1073973	0.0	1.200	1499.2
SYDNEY	2008	2.9579235	0.0	1.800	1082.6
SYDNEY	2009	2.6197260	0.0	0.800	956.2
SYDNEY	2010	3.1878453	0.0	1.800	1154.0
SYDNEY	2011	3.7512329	0.0	2.000	1369.2
SYDNEY	2012	3.3158470	0.0	1.200	1213.6
SYDNEY	2013	3.6832877	0.0	0.600	1344.4
SYDNEY	2014	2.4853186	0.0	1.600	897.2
SYDNEY	2015	3.6663014	0.0	1.400	1338.2
SYDNEY	2016	3.7857923	0.0	0.750	1385.6
SYDNEY	2017	2.9630854	0.0	0.700	1075.6
SYDNEY	2018	2.7478022	0.0	0.800	1000.2
SYDNEY	2019	2.3336986	0.0	0.200	851.8

Insights from Datasets:

• Upward Temperature Trend: There is a noticeable upward trend in temperatures, particularly in the later years of the dataset, with several cities showing higher average and median temperatures. This is consistent with the global trend of rising temperatures and could be indicative of climate change effects.

• Seasonal Variability: The wide range between the minimum and maximum temperatures for each city within each year indicates strong seasonal variability. This is expected, but the range also shows the potential for extreme weather events, as seen with some very high maximum temperatures.

• Rainfall Variability: The rainfall data exhibits considerable yearly variability within cities, with certain years showing peaks that could correspond to specific heavy rainfall events. This suggests that rainfall is not consistent year-to-year and is subject to larger climatic patterns or events.

• Data Quality: Initially, the median rainfall values being zero indicated potential data issues or many zero-rainfall days. However, the 75th percentile values showed that significant rainfall did occur, indicating the data likely captured actual variations in rainfall.

• Extreme Values: The dataset contains some extreme temperature values, both hot and cold, which could warrant further investigation to understand if they are outliers, errors, or true weather phenomena.

• Consistency Across Years: While there is year-to-year variation, the temperature data generally show a consistent pattern over the two-decade span, with the recent years standing out for their higher values.

Time, Date, and Timeseries

Time Conversion

Timeseries is all about a repeating trend. Sometimes you don’t know that there is a recurring trend until you create the data as a timeseries. Here, we’ll go ahead and use ts() on the variables that, in your opinion, it would make sense to apply it to.

The frequency argument inside ts() dictates how many observations should be considered in each cycle of the recurring trend.

• weekly, monthly, yearly, etc.

library(lubridate)
library(zoo)

rainfall$date <- as.Date(paste(rainfall$year, rainfall$month, rainfall$day, sep = "-"), "%Y-%m-%d")
temperature$date <- as.Date(temperature$date)

rainfall_filtered <- subset(rainfall, year >= 2000 & (year < 2019 | (year == 2019 & as.Date(paste(year, month, day, sep="-")) <= as.Date("2019-05-31"))))
temperature_filtered <- subset(temperature, year >= 2000 & (year < 2019 | (year == 2019 & date <= as.Date("2019-05-31"))))

rainfall_filtered %>% 
  summarise(across(everything(), ~sum(is.na(.))))

## # A tibble: 1 × 12
##   station_code city_name  year month   day rainfall period quality   lat  long
##          <int>     <int> <int> <int> <int>    <int>  <int>   <int> <int> <int>
## 1            0         0     0     0     0     3160  18974    3164     0     0
## # ℹ 2 more variables: station_name <int>, date <int>

temperature_filtered %>% 
  summarise(across(everything(), ~sum(is.na(.))))

## # A tibble: 1 × 7
##   city_name  date temperature temp_type site_name  year month
##       <int> <int>       <int>     <int>     <int> <int> <int>
## 1         0     0          34         0         0     0     0

numeric_columns <- sapply(rainfall_filtered, is.numeric)
rainfall_filtered[numeric_columns] <- lapply(rainfall_filtered[numeric_columns], function(x) {ifelse(is.na(x), 0, x)})

temperature_filtered <- na.omit(temperature_filtered)

yearly_rainfall_by_city <- aggregate(rainfall ~ city_name + year, data = rainfall_filtered, sum)
yearly_temperature_by_city <- aggregate(temperature ~ city_name + year, data = temperature_filtered, mean)

monthly_rainfall_by_city <- aggregate(rainfall ~ city_name + year + month, data = rainfall_filtered, sum)
monthly_temperature_by_city <- aggregate(temperature ~ city_name + year + month, data = temperature_filtered, mean)

rainfall_list <- split(yearly_rainfall_by_city, yearly_rainfall_by_city$city_name)
temperature_list <- split(yearly_temperature_by_city, yearly_temperature_by_city$city_name)


rainfall_ts_list <- lapply(rainfall_list, function(df) {
    start_year <- ifelse(df$city_name[1] == "Canberra", 2008, min(df$year))
    end_year <- max(df$year)
    ts(df$rainfall, start=start_year, end=end_year, frequency=1)
})


temperature_ts_list <- lapply(temperature_list, function(df) {
    ts(df$temperature, start=min(df$year), end=max(df$year), frequency=1)
})

monthly_rainfall_list <- split(monthly_rainfall_by_city, monthly_rainfall_by_city$city_name)
monthly_temperature_list <- split(monthly_temperature_by_city, monthly_temperature_by_city$city_name)


monthly_rainfall_ts_list <- lapply(monthly_rainfall_list, function(df) {
    start_year <- ifelse(df$city_name[1] == "Canberra", 2008, min(df$year))
    ts(df$rainfall, start = c(start_year, min(df$month[df$year == start_year])), frequency = 12)
})

monthly_temperature_ts_list <- lapply(monthly_temperature_list, function(df) {
    ts(df$temperature, start = c(min(df$year), min(df$month)), frequency = 12)
})

# Function to convert a time series object to a data frame
convert_ts_to_df <- function(ts_obj, city_name, value_name) {
  years <- as.integer(time(ts_obj))
  values <- as.vector(ts_obj)
  data.frame(city_name = city_name, year = years, value = values, stringsAsFactors = FALSE)
}

# Convert each time series in the list to a data frame and then combine them
rainfall_df <- do.call(rbind, lapply(names(rainfall_ts_list), function(city) {
  convert_ts_to_df(rainfall_ts_list[[city]], city, "rainfall")
}))

temperature_df <- do.call(rbind, lapply(names(temperature_ts_list), function(city) {
  convert_ts_to_df(temperature_ts_list[[city]], city, "temperature")
}))

library(stats)
library(tseries)
library(forecast)
library(tidyr)



adf_test_results_monthly_temperature <- lapply(monthly_temperature_ts_list, adf.test)
adf_test_results_monthly_temperature

## $BRISBANE
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -0.33177, Lag order = 6, p-value = 0.9889
## alternative hypothesis: stationary
## 
## 
## $CANBERRA
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -0.05211, Lag order = 6, p-value = 0.99
## alternative hypothesis: stationary
## 
## 
## $MELBOURNE
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -0.056618, Lag order = 6, p-value = 0.99
## alternative hypothesis: stationary
## 
## 
## $PERTH
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -0.067175, Lag order = 6, p-value = 0.99
## alternative hypothesis: stationary
## 
## 
## $SYDNEY
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -0.16611, Lag order = 6, p-value = 0.99
## alternative hypothesis: stationary

adf_test_results_monthly_rainfall <- lapply(monthly_rainfall_ts_list, adf.test)
adf_test_results_monthly_rainfall

## $BRISBANE
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -3.7515, Lag order = 6, p-value = 0.02218
## alternative hypothesis: stationary
## 
## 
## $CANBERRA
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -4.0377, Lag order = 5, p-value = 0.01
## alternative hypothesis: stationary
## 
## 
## $MELBOURNE
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -6.1863, Lag order = 6, p-value = 0.01
## alternative hypothesis: stationary
## 
## 
## $PERTH
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -1.6024, Lag order = 6, p-value = 0.743
## alternative hypothesis: stationary
## 
## 
## $SYDNEY
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -4.8967, Lag order = 6, p-value = 0.01
## alternative hypothesis: stationary

adf_test_results_yearly_rainfall <- lapply(rainfall_ts_list, adf.test)
adf_test_results_yearly_rainfall

## $BRISBANE
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -0.65474, Lag order = 2, p-value = 0.9621
## alternative hypothesis: stationary
## 
## 
## $CANBERRA
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -1.2307, Lag order = 2, p-value = 0.8655
## alternative hypothesis: stationary
## 
## 
## $MELBOURNE
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -1.5964, Lag order = 2, p-value = 0.7261
## alternative hypothesis: stationary
## 
## 
## $PERTH
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -3.7968, Lag order = 2, p-value = 0.03594
## alternative hypothesis: stationary
## 
## 
## $SYDNEY
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -1.4466, Lag order = 2, p-value = 0.7832
## alternative hypothesis: stationary

adf_test_results_yearly_temperature <- lapply(temperature_ts_list, adf.test)
adf_test_results_yearly_temperature

## $BRISBANE
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = 0.10806, Lag order = 2, p-value = 0.99
## alternative hypothesis: stationary
## 
## 
## $CANBERRA
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = 0.29541, Lag order = 2, p-value = 0.99
## alternative hypothesis: stationary
## 
## 
## $MELBOURNE
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -1.3562, Lag order = 2, p-value = 0.8177
## alternative hypothesis: stationary
## 
## 
## $PERTH
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = -2.9611, Lag order = 2, p-value = 0.2063
## alternative hypothesis: stationary
## 
## 
## $SYDNEY
## 
##  Augmented Dickey-Fuller Test
## 
## data:  X[[i]]
## Dickey-Fuller = 0.14916, Lag order = 2, p-value = 0.99
## alternative hypothesis: stationary

Which relationships are evident? Are they a long-term trend or do they just exhibit seasonality?

• Monthly Temperature Dataset (First Set of Results): Most cities (like Brisbane, Canberra, Kent, Melbourne) have high p-values (close to 0.99), suggesting non-stationarity, indicating the presence of trends in the monthly temperature data.

• Monthly Rainfall Dataset (Second Set of Results): Some cities (like Brisbane, Canberra, Melbourne, Sydney) show low p-values (e.g., < 0.05), indicating stationarity and thus no significant trend. Other cities show high p-values, suggesting non-stationarity and potential trends.

• Yearly Rainfall Dataset (Third Set of Results): Cities like Perth show low p-values, indicating stationarity and no significant long-term trend.Other cities have higher p-values, suggesting potential trends.

• Yearly Temperature Dataset (Fourth Set of Results): Cities like Brisbane and Canberra show high p-values, indicating non-stationarity and possible trends. Other cities, like Perth, show lower p-values, suggesting no significant long-term trend.

Plotting

common_cities <- intersect(names(temperature_ts_list), names(rainfall_ts_list))

# Rainfall plot
rainfall_df_common <- do.call(rbind, lapply(common_cities, function(city) {
  convert_ts_to_df(rainfall_ts_list[[city]], city, "rainfall")
}))

ggplot(rainfall_df_common, aes(x = year, y = value, group = city_name, color = city_name)) +
  geom_line(lwd=1.1) +
  facet_wrap(~ city_name, scales = "free_x") +
  labs(title = "Yearly Rainfall by City (2000-2019)", x = "Year", y = "Total Rainfall") +
  first_custom() +
  theme(legend.position = "none") 

# Temperature plot
temperature_df_common <- do.call(rbind, lapply(common_cities, function(city) {
  convert_ts_to_df(temperature_ts_list[[city]], city, "temperature")
}))

ggplot(temperature_df_common, aes(x = year, y = value, group = city_name, color = city_name)) +
  geom_line(lwd=1.1) +
  facet_wrap(~ city_name, scales = "free_x") +
  labs(title = "Yearly Mean Temperature by City (2000-2019)", x = "Year", y = "Mean Temperature") +
  first_custom() +
  theme(legend.position = "none")

The plots of temperature and rainfall trends by city from 2000 to 2019 show:

Seasonal patterns are evident in both temperature and rainfall data. There’s a slight upward trend in temperatures for most cities, suggesting a warming pattern. Rainfall data is more variable with no clear long-term trend.

Predicting 100 years into the future is complex, but if current warming trends persist, temperatures are likely to continue rising. Rainfall predictions are less certain due to their variability.

Geospatial

Plotting

Geospatial data can get very complicated very quickly with

• different coordinate systems
• data formats like shapefile, raster, tiff
• data grouping
• and many different specialized packages for handling this data.

There are powerful applications like ArcGIS that specialize in processing this data. But we will be just fine with ggplot2. This week’s TidyTuesday project happens to involve geospatial data!

• map_data()
  • Resource
• geom_sf()
• group =
• coord_quickmap()

library(sf)
library(raster)
library(dplyr)
library(ggrepel)
library(viridis)
library(ggplot2)
library(rnaturalearth)
library(rnaturalearthhires)
# Load Australia shapefile

australia <- rnaturalearth::ne_states(geounit = "australia")
sp::plot(australia)

# Load and plot raster data
grid_raster <- raster("2019120120191231.grid")

plot(grid_raster)

climate <- as(grid_raster, "SpatialPixelsDataFrame")
climate <- as.data.frame(climate)
climate <- climate %>%
  rename(temp = X2019120120191231)

# Select and clean Australia cities data
australia_cities <- rainfall %>%
  dplyr::select(city_name, long, lat) %>%
  distinct()
australia_cities <- australia_cities[-4,]

# Create the plot
tempdec <- ggplot() +
  geom_tile(data = climate, aes(x = x, y = y, fill = temp)) +
  geom_sf(data = australia, fill = NA, color = "#3E3E3E", size = 0.1) +
  geom_point(data = australia_cities, aes(x = long, y = lat), color = "white") +
  geom_text_repel(data = australia_cities, aes(x = long, y = lat, label = city_name), size = 3, color = "white") +
  scale_fill_viridis_c(option = "B") +
  coord_sf(crs = st_crs(australia)) +
  theme_void() +
  theme(
    panel.background = element_blank(),
    panel.grid.major = element_blank(),
    panel.grid.minor = element_blank(),
    legend.background = element_blank(),
    axis.text = element_blank(),
    axis.title = element_blank(),
    plot.caption = element_text(vjust = 1.1, hjust = 1.3, size = 7),
    legend.position = "right",
    legend.justification = "center",
    legend.box.just = "right"
  ) +
  labs(
    title = "Australia Climate",
    subtitle = "Average High Temperatures for December 2019",
    caption = "Australian Bureau of Meteorology"
  )

# Plot the result
print(tempdec)

The plot is a thematic map depicting the average high temperatures across Australia for December 2019. It employs a gradient color scheme, where warmer colors represent higher temperatures, creating a visual representation of the heat distribution. Major cities are marked with white points, and labels provide clear identification, adding a spatial context to the temperature data.

---

Links

• https://www.r-graph-gallery.com/time-series.html
• https://www.r-graph-gallery.com/line_chart_annotation.html
• https://www.r-graph-gallery.com/line-chart-dual-Y-axis-ggplot2.html
• https://fred.stlouisfed.org/
• http://austinwehrwein.com/