Arizona is known for its extreme temperatures, with prolonged periods of high heat posing serious risks to the environment, agriculture, and public health. In particular, rising temperatures have been closely linked to increased drought frequency and severity across the state. Understanding when and where these temperature extremes occur is critical to identifying patterns that may contribute to drought conditions.

Our mission is to analyze historical temperature data across Arizona counties to identify seasonal and regional patterns in extreme heat. By visualizing these trends, we aim to support early detection of conditions that may lead to drought, and contribute insights that can assist in climate resilience planning and environmental preparedness.

This project was a team effort. Tasks were divided as follows:

• Bhavani Gali – Background, problem definition, and data wrangling
• Gabriela Hinostroza – Exploratory data analysis and visualizations
• Sai Aishwarya – Regression modeling and final reporting

We used a dataset titled “US Weekly Temperature by County”, which contains weekly temperature data across U.S. counties, including those in Arizona.

• Source: Kaggle - US Weekly Temperature by County (https://www.kaggle.com/datasets/akhilv11/us-weekly-temperature-by-county)

We filtered the data to include only counties within Arizona, enabling us to focus our analysis on regional climate trends and temperature patterns that may relate to drought conditions.

• Variables Included in Our Subset:

  • date: Calendar date of the observation

  • Temperature_F: Temperature in degrees Fahrenheit

  • week_num: Week number of the year (1–52)

  • year: Extracted from the date

  • Month: Month name derived from the date

df <- read_csv("cleaned_arizona_temperature.csv")
df <- df %>%
  mutate(Month = month(date, label = TRUE)) 

This histogram displays the frequency distribution of daily temperatures across Arizona. Most temperatures fall between 40°F and 80°F, indicating a wide range of mild to hot days, with fewer occurrences of extreme cold or extreme heat.

hist_plot <- ggplot(df, aes(x = Temperature_F)) +
  geom_histogram(binwidth = 2, fill = "skyblue", color = "black") +
  labs(x = "Temperature (°F)",
       y = "Frequency")

This boxplot shows the variation in temperatures by month. It reveals that summer months like June through August have the highest median temperatures, while winter months like December and January have the lowest. Some outliers may indicate unusually hot or cold days.

box_plot <- ggplot(df, aes(x = Month, y = Temperature_F)) +
  geom_boxplot(fill = "lightgreen") +
  labs(x = "Month",
       y = "Temperature (°F)")

This line plot shows how temperatures have changed day by day over the full date range. The seasonal cycle is clear, with temperature rising and falling regularly each year, providing a good visual of long-term temperature fluctuation in Arizona.

time_plot <- ggplot(df, aes(x = date, y = Temperature_F)) +
  geom_line(color = "steelblue") +
  labs(x = "Date",
       y = "Temperature (°F)")

This scatterplot visualizes how temperature changes by week of the year. The U-shaped trend line shows that temperatures are lowest around January (weeks 1–5) and highest in summer (weeks 25–35), which matches Arizona’s known climate pattern.

p <- ggplot(df, aes(x = week_num, y = Temperature_F)) +
  geom_point(alpha = 0.5, color = "darkorange") +   
  geom_smooth(method = "lm", color = "blue", se = FALSE) +
  theme_minimal() +
  labs(x = "Week Number", y = "Temperature (°F)")

This linear regression model predicts temperature using both the week number and the year.

df <- df %>% mutate(yr = lubridate::year(date))

temp_model <- lm(Temperature_F ~ week_num + yr, data = df)

This table shows the estimated effect of week number and year on temperature in Fahrenheit.

summary(temp_model)$coefficients

##                 Estimate   Std. Error    t value     Pr(>|t|)
## (Intercept) -11.43226983 112.49803242 -0.1016220 9.190591e-01
## week_num      0.19688906   0.01174598 16.7622536 4.452140e-62
## yr            0.03488791   0.05574701  0.6258256 5.314460e-01

This plot shows yearly temperature patterns by week in Arizona. Each line represents a different year. While the U-shaped pattern is consistent, later years show a slight upward shift, supporting a gradual warming trend.

r1 <- ggplot(df, aes(x = week_num, y = Temperature_F, color = factor(yr))) +
  geom_point(alpha = 0.4, size = 0.7) +
  geom_smooth(method = "lm", se = FALSE, linewidth = 0.8) +
  theme_minimal() +
  labs(x = "Week Number", 
       y = "Temperature (°F)", 
       color = "Year")

Through our analysis of Arizona’s historical temperature data, we observed clear seasonal patterns, with the hottest weeks occurring during the summer months and cooler temperatures in winter. Our regression model confirmed that week number is a strong predictor of temperature, while year has a weaker but slightly positive effect, suggesting a gradual warming trend. These findings support the relevance of temperature trends in understanding potential drought conditions and can aid in planning for climate resilience efforts across the state.