计算满足条件的连续天数的平均值

Question

I'm using a cods where I can detect heat wave for consecutive days, but I cannot figure how I can also get the cods select the highest mean temperature recorded over every heat weave event. The header of my data (Period1): Year Month Day Mean MAX MIN

My codes are:

setDT(Period1)
Period1[, hotday := +(Period1$MAX>=(quantile(Period1$MAX,.9, na.rm = T, type = 6)) & Period1$MIN>=(quantile(Period1$MIN,.8, na.rm = T, type = 6)))
        ][, hw.length := with(rle(hotday), rep(lengths,lengths))
          ][hotday==0, hw.length:=0]

The result of the code:

YEAR    MONTH   DAY MEAN    MAX MIN hotday  hw.length
1   2005    7   1   34.7    42   28     0   0
2   2005    7   2   35     41.8  28.8   0   0
3   2005    7   3   34.6   41.5  27     1   2
4   2005    7   4   35.4   43   27.6    1   2
5   2005    7   5   35.4   42   27.4    0   0
6   2005    7   6   34.4   42.2 27.7    0   0
7   2005    7   7   34     39.3 28.6    1   4
8   2005    7   8   34.9   40.6  29     1   4
9   2005    7   9   35.7   41  28.2     1   4
10  2005    7   10  35.1   42.2 28.5    1   4
11  2005    7   11  34.9   42.5  29     0   0
12  2005    7   12  35.5   43.2  28     0   0
13  2005    7   13  36.1   43  29.2     1   3
14  2005    7   14  36.4   43.4  32     1   3
15  2005    7   15  36.5   44.5  29.2   1   3
16  2005    7   16  36.2   42.6  31.2   0   0
17  2005    7   17  34.6   40    30     0   0
18  2005    7   18  33.7   41  28.8     0   0

For heat wave on 2005/7/3 and 2005/7/4 the highest is 35.6 , where for the second heat wave that stared on 2005/7/7 and ended on 2005/7/10 the highest mean was 35.7 . I would appreciate any idea on how I can include extracting the highest Mean of every heat wave considering the consecutive days.

Answer 1

With:

Period1[hotday == 1, maxmean := max(MEAN) , rleid(hw.length)][]

you get:

    YEAR MONTH DAY MEAN  MAX  MIN hotday hw.length maxmean
 1: 2005     7   1 34.7 42.0 28.0      0         0      NA
 2: 2005     7   2 35.0 41.8 28.8      0         0      NA
 3: 2005     7   3 34.6 41.5 27.0      1         2    35.4
 4: 2005     7   4 35.4 43.0 27.6      1         2    35.4
 5: 2005     7   5 35.4 42.0 27.4      0         0      NA
 6: 2005     7   6 34.4 42.2 27.7      0         0      NA
 7: 2005     7   7 34.0 39.3 28.6      1         4    35.7
 8: 2005     7   8 34.9 40.6 29.0      1         4    35.7
 9: 2005     7   9 35.7 41.0 28.2      1         4    35.7
10: 2005     7  10 35.1 42.2 28.5      1         4    35.7
11: 2005     7  11 34.9 42.5 29.0      0         0      NA
12: 2005     7  12 35.5 43.2 28.0      0         0      NA
13: 2005     7  13 36.1 43.0 29.2      1         3    36.5
14: 2005     7  14 36.4 43.4 32.0      1         3    36.5
15: 2005     7  15 36.5 44.5 29.2      1         3    36.5
16: 2005     7  16 36.2 42.6 31.2      0         0      NA
17: 2005     7  17 34.6 40.0 30.0      0         0      NA
18: 2005     7  18 33.7 41.0 28.8      0         0      NA

Explanation :

• Filter the data for only the hot days: hotday == 1 (or with some code-golfing: !!hotday ).
• For the remaining rows, create a runlength id with the rleid -function so that you can group by heatwave: rleid(hw.length) .
• Finally extract the maximum mean for each heat wave and assign it to a new colum with: maxmean := max(MEAN) .

---

If you just want to extract the maximum values for the heat waves, you can use:

> Period1[!!hotday, max(MEAN) , rleid(hw.length)]$V1
[1] 35.4 35.7 36.5

---

Used data:

Period1 <- fread('YEAR    MONTH   DAY MEAN    MAX MIN hotday  hw.length
2005    7   1   34.7    42   28     0   0
2005    7   2   35     41.8  28.8   0   0
2005    7   3   34.6   41.5  27     1   2
2005    7   4   35.4   43   27.6    1   2
2005    7   5   35.4   42   27.4    0   0
2005    7   6   34.4   42.2 27.7    0   0
2005    7   7   34     39.3 28.6    1   4
2005    7   8   34.9   40.6  29     1   4
2005    7   9   35.7   41  28.2     1   4
2005    7   10  35.1   42.2 28.5    1   4
2005    7   11  34.9   42.5  29     0   0
2005    7   12  35.5   43.2  28     0   0
2005    7   13  36.1   43  29.2     1   3
2005    7   14  36.4   43.4  32     1   3
2005    7   15  36.5   44.5  29.2   1   3
2005    7   16  36.2   42.6  31.2   0   0
2005    7   17  34.6   40    30     0   0
2005    7   18  33.7   41  28.8     0   0')

Answer 2

I don't know exactly how your code in details look like, but here is a hint assuming Period1 is a dataframe:

Extract the hot period

hotPeriod <- Period1[, hotday := +(Period1$MAX>=(quantile(Period1$MAX,.9, na.rm = T, type = 6)) & Period1$MIN>=(quantile(Period1$MIN,.8, na.rm = T, type = 6)))
  ][, hw.length := with(rle(hotday), rep(lengths,lengths))
    ][hotday==0, hw.length:=0]

Extract the maximum of the mean value over that period

max.of.mean <- max( hotPeriod$Mean )