CODEWITHSUNDEEP
rdataframedateggplot2dplyr
twseewx
twseewx

Reputation: 332

How can I plot hourly averaged time series grouped by month?

I have a long term time series (22 years), and want to group the data by month and average each hour to get a diurnal temperature profile. Below is my data set.

        Datetime.LST ruc.197.118 narr.225.118 KLAF VAIv212_08748740771
1   12/31/1996 18:00          NA      -0.7843   NA                  NA
2   12/31/1996 19:00          NA      -1.2432   NA                  NA
3   12/31/1996 20:00          NA      -1.7022   NA                  NA
4   12/31/1996 21:00          NA      -2.1611   NA                  NA
5   12/31/1996 22:00          NA      -2.1026   NA                  NA
6   12/31/1996 23:00          NA      -2.0440   NA                  NA
7      1/1/1997 0:00          NA      -1.9854   NA           0.4120480
8      1/1/1997 1:00          NA      -1.7506   NA           0.4400940
9      1/1/1997 2:00          NA      -1.5157   NA           0.4594420
10     1/1/1997 3:00          NA      -1.2808   NA           0.3232730
11     1/1/1997 4:00          NA      -0.9326   NA           0.0754700
12     1/1/1997 5:00          NA      -0.5844   NA          -0.0730896
13     1/1/1997 6:00          NA      -0.2363   NA          -0.1574400
14     1/1/1997 7:00          NA       0.3422   NA          -0.2461240
15     1/1/1997 8:00          NA       0.9207   NA          -0.0277405

  temperature_data <- read.csv(temp_hourly, stringsAsFactors = FALSE)
  monthly_hourly <- temperature_data %>%
              dmy_hm(.$Datetime.LST, tz = "GMT") %>%
              dplyr::group_by(month=format(Datetime.LST,'%m'), hour=format(as.POSIXct(cut(Datetime.LST, breaks = 'hour')),"%H")) %>%
              summarise(meanVal=mean(val,na.rm=TRUE))
}

Example dataset using dput

> dput(head(temperature_data[, c(1,2,3,4,5)]))
structure(list(Datetime.LST = c("12/31/1996 18:00", "12/31/1996 19:00", 
"12/31/1996 20:00", "12/31/1996 21:00", "12/31/1996 22:00", "12/31/1996 23:00"
), ruc.197.118 = c(NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, 
NA_real_), narr.225.118 = c(-0.7843, -1.2432, -1.7022, -2.1611, 
-2.1026, -2.044), KLAF = c(NA_integer_, NA_integer_, NA_integer_, 
NA_integer_, NA_integer_, NA_integer_), VAIv212_08748740771 = c(NA_real_, 
NA_real_, NA_real_, NA_real_, NA_real_, NA_real_)), .Names = c("Datetime.LST", 
"ruc.197.118", "narr.225.118", "KLAF", "VAIv212_08748740771"), row.names = c(NA, 
6L), class = "data.frame")

Datetime.LST is brought in as ch data type. I have had some issues getting different date packages(lubridate) to agree with dplyr's group_by function.

Attached is an example plot output that I would like to see. Monthly_Diurnal_Temperature_Profiles

With using Andrews comments below, updated code and current output:

  temperature_data <- read.csv(temp_hourly, stringsAsFactors = FALSE)
  monthly_hourly <-   temperature_data %>%
    mutate(Datetime.LST = mdy_hm(Datetime.LST, tz = "GMT"),
           day = day(Datetime.LST),
           month = month(Datetime.LST),
           hour = hour(Datetime.LST)) %>%
    group_by(month, hour) %>% 
    summarise(
      ruc.197.118 = mean(ruc.197.118, na.rm = TRUE),
      narr.225.118 = mean(narr.225.118, na.rm = TRUE),
      KLAF = mean(KLAF, na.rm = TRUE),
      VAIv212_08748740771 = mean(VAIv212_08748740771, na.rm = TRUE)
    ) %>%
    # next step creates a dummy year to group month/day/hr as datetime
    mutate(month_name = month(month, label = TRUE)) 
  ggplot(data = monthly_hourly,aes(x = hour)) +
  geom_line(color = "red", aes(y = ruc.197.118)) + 
  geom_line(color = "blue", aes(y = narr.225.118)) + 
  geom_line(color = "black", aes(y = KLAF)) + 
  geom_line(color = "orange", aes(y = VAIv212_08748740771)) + 
  xlab("Temperature degC") +
  ylab("Hour of Day") +
  facet_wrap(~ month_name)

newoutput.

Upvotes: 0

Views: 722

Answers (1)

Andy Baxter
Andy Baxter

Reputation: 7626

Following on from your code above, it would work to extract date, month and hour as new variables, then group, summarise and graph:

monthly_hourly <- df %>%
  bind_rows(data.frame(Datetime.LST = "1/1/1970 1:00")) %>% 
  mutate(Datetime.LST = mdy_hm(Datetime.LST, tz = "GMT"),
         day = day(Datetime.LST),
         month = month(Datetime.LST, label = TRUE),
         hour = hour(Datetime.LST)) %>%
  group_by(month, hour) %>% 
  summarise(
    ruc.197.118 = mean(ruc.197.118, na.rm = TRUE),
    narr.225.118 = mean(narr.225.118, na.rm = TRUE),
    KLAF = mean(KLAF, na.rm = TRUE),
    VAIv212_08748740771 = mean(VAIv212_08748740771, na.rm = TRUE)
  )

monthly_hourly %>% 
  ggplot(aes(hour)) +
  geom_line(color = "red", aes(y = ruc.197.118)) + 
  geom_line(color = "blue", aes(y = narr.225.118)) + 
  geom_line(color = "black", aes(y = KLAF)) + 
  geom_line(color = "orange", aes(y = VAIv212_08748740771)) + 
  facet_wrap(~ month, scales = "free")

Upvotes: 0

Related Questions