plotting julian dates across 2 calendar years

Question

I'm sure this will be flagged as duplicate but I cannot seem to find the solution.

I have a dataset consisting of nitrogen measured in the atmosphere across time. Measurements were collected daily from November to February of each year (~20 years). The dates are in Julian format. Default specs in ggplot2 treats Feb as the first month which should be Nov. How can I address this while keeping in mind the x-axis should be true (e.g., Nov - Dec - Jan - Feb). Thanks in advance.

structure(list(nitro = c(36, 51, 23, 7, 26, 62, 45, 25, 31, 70, 
23, 63, 41, 61, 6, 24, 2, 6, 47, 22, 6, 48, 16, 25, 61, 9, 63, 
3, 22, 27, 52, 9, 34, 48, 31, 16, 51, 63, 21, 42, 4, 9, 20, 28, 
23, 0, 23, 70, 47, 2, 52.0898926784594, 55.8022284640568, 41.3593089128863, 
57.9928202206287, 35.4624211011211, 65.2484486428607, 39.236626265918, 
40.8906848288849, 65.7044590718657, 39.9377208120021, 51.1368297792738, 
46.3981830973577, 36.3795615108741, 59.4071137999101, 42.77053802297, 
51.4293854584156, 45.3589706700315, 53.9652146584217, 54.6980728446628, 
43.1278297603877, 48.2879709803009, 60.0627232974633, 49.5233404192228, 
55.4231572422576, 35.9842318889366, 55.649768270307, 61.8309485366259, 
63.1600858734154, 52.9116478602541, 48.5690637010561, 49.4065375720883, 
44.7880773761964, 48.5893427534388, 43.5882970429564, 33.0365374509187, 
40.2149387649389, 67.3129711280128, 55.491540487927, 63.5614232930341, 
53.6421935293565, 47.0292298358771, 62.5727076637489, 32.2728901281264, 
61.7219468394218, 40.0865833610807, 24.1288862068719, 62.8734959914056, 
52.3161174228334, 32.6827415750796, 50.3593375610126), juliandate = c(323L, 
346L, 318L, 314L, 324L, 325L, 311L, 342L, 341L, 348L, 328L, 345L, 
328L, 328L, 326L, 312L, 340L, 331L, 326L, 320L, 359L, 323L, 312L, 
337L, 347L, 351L, 352L, 320L, 326L, 345L, 352L, 342L, 312L, 323L, 
333L, 348L, 340L, 330L, 338L, 323L, 356L, 345L, 359L, 334L, 320L, 
349L, 358L, 355L, 321L, 324L, 6L, 52L, 46L, 21L, 47L, 1L, 43L, 
14L, 48L, 2L, 60L, 21L, 27L, 26L, 37L, 51L, 2L, 1L, 43L, 57L, 
23L, 49L, 10L, 38L, 58L, 20L, 3L, 45L, 1L, 17L, 30L, 42L, 5L, 
58L, 1L, 4L, 7L, 38L, 56L, 53L, 14L, 46L, 54L, 36L, 52L, 6L, 
32L, 35L, 46L, 3L)), .Names = c("nitro", "juliandate"), row.names = c(NA, 
100L), class = "data.frame")

ggplot(d,aes(juliandate,nitro))+geom_point() #plot - should start at julian 305, end at julian 60

Try adding the code you tried so far, your best shot so-to-say. Not doing that is probably the reason why you are getting downvoted, "no code" questions about code are not treated kindly here. — Mike Wise, Jan 16 '16 at 22:44

Mike Wise · Accepted Answer · 2016-01-19T22:38:03.333

The best solution I see is to map them all to a single year and plot that year using ggplot2 functionality to make the date look like a Julian date.

So there are two parts to the solution, a mapping, and the display. In your application I would realize these as separate columns in a dataframe.

The solution presented below uses a wrapper function (j2d)for the mapping that wraps the date at mid-year (180 days). It then uses the ggplot2 function scale_x_date to convert the date on output to a Julian format:

library(ggplot2)

j2d <- function(n){
  n <- ifelse( n<180, n+365, n ) # wrap at midyear
  d <- as.Date("2015-01-01") + n
  return(d)
}

df <-
  structure(
    list( 
      nitro = 
        c(36, 51, 23, 7, 26, 62, 45, 25, 31, 70, 
          23, 63, 41, 61, 6, 24, 2, 6, 47, 22, 6, 48, 16, 25, 61, 9, 63, 
          3, 22, 27, 52, 9, 34, 48, 31, 16, 51, 63, 21, 42, 4, 9, 20, 28, 
          23, 0, 23, 70, 47, 2, 52.0898926784594, 55.8022284640568, 41.3593089128863, 
          57.9928202206287, 35.4624211011211, 65.2484486428607, 39.236626265918, 
          40.8906848288849, 65.7044590718657, 39.9377208120021, 51.1368297792738, 
          46.3981830973577, 36.3795615108741, 59.4071137999101, 42.77053802297, 
          51.4293854584156, 45.3589706700315, 53.9652146584217, 54.6980728446628, 
          43.1278297603877, 48.2879709803009, 60.0627232974633, 49.5233404192228, 
          55.4231572422576, 35.9842318889366, 55.649768270307, 61.8309485366259, 
          63.1600858734154, 52.9116478602541, 48.5690637010561, 49.4065375720883, 
          44.7880773761964, 48.5893427534388, 43.5882970429564, 33.0365374509187, 
          40.2149387649389, 67.3129711280128, 55.491540487927, 63.5614232930341, 
          53.6421935293565, 47.0292298358771, 62.5727076637489, 32.2728901281264, 
          61.7219468394218, 40.0865833610807, 24.1288862068719, 62.8734959914056, 
          52.3161174228334, 32.6827415750796, 50.3593375610126), 
      juliandate = 
        c(323L, 
          346L, 318L, 314L, 324L, 325L, 311L, 342L, 341L, 348L, 328L, 345L, 
          328L, 328L, 326L, 312L, 340L, 331L, 326L, 320L, 359L, 323L, 312L, 
          337L, 347L, 351L, 352L, 320L, 326L, 345L, 352L, 342L, 312L, 323L, 
          333L, 348L, 340L, 330L, 338L, 323L, 356L, 345L, 359L, 334L, 320L, 
          349L, 358L, 355L, 321L, 324L, 6L, 52L, 46L, 21L, 47L, 1L, 43L, 
          14L, 48L, 2L, 60L, 21L, 27L, 26L, 37L, 51L, 2L, 1L, 43L, 57L, 
          23L, 49L, 10L, 38L, 58L, 20L, 3L, 45L, 1L, 17L, 30L, 42L, 5L, 
          58L, 1L, 4L, 7L, 38L, 56L, 53L, 14L, 46L, 54L, 36L, 52L, 6L, 
          32L, 35L, 46L, 3L)), 
    .Names = c("nitro", "juliandate"), 
    row.names = c(NA, 100L), class = "data.frame")

df$juliandate<-j2d(df$juliandate)

ggplot(df,aes(juliandate,nitro))+
  geom_point() +
  scale_x_date(labels = date_format("%j"))

Yielding:

Mike - I appreciate your time but I was not clear with my previous coding - I didn't want your function applied towards the codes that created the arbitrary df but wanted the function applied towards the df itself. Does that make sense? — ecologist55, Jan 19 '16 at 19:30
That is accommodated with a small change to the example code. Is this what you are looking for? If not you will need to be more clear. — Mike Wise, Jan 19 '16 at 19:46
Perfect! I'll amend the example dataset so that others can use if necessary. Thanks Mike. — ecologist55, Jan 19 '16 at 21:26
If you are happy, mark it as correct then please. I will put the new data in my version too. — Mike Wise, Jan 19 '16 at 21:31

plotting julian dates across 2 calendar years

1 Answers1