Daily minimum values in R

Question

I am trying to extract the daily minimum zenith angle in a dataset which consists of 24h values (1 zenith angle value every hour) over ~31 days for 12 months. It looks like this:

JulianDay Azimuth Zenith Date (YYMMDD HH:MM:SS)
2455928 174.14066 70.04650 2012-01-01 13:00:00
2455928 188.80626 70.30747 2012-01-01 14:00:00
2455928 203.03458 73.12297 2012-01-01 15:00:00
2455928 216.28061 78.20131 2012-01-01 16:00:00
2455928 228.35929 85.10759 2012-01-01 17:00:00
....
2456293 146.33844 77.03456 2012-12-31 11:00:00
2456293 159.80472 72.38003 2012-12-31 12:00:00

Is there a function that can extract the maximum and minimum solar zenith angle from each day (i.e., 365 outputs)?

Answer 1

You can do a summary grouped by day, here is one way, suppose your data frame is called df:

library(data.table)
setDT(df)[, .(maxZenith = max(Zenith), minZenith = min(Zenith)), .(JulianDay)]

If you want to use the Date column instead of JulianDay, do something like:

setDT(df)[, .(maxZenith = max(Zenith), minZenith = min(Zenith)), .(as.Date(Date))]

Assuming you renamed your Date (YYMMDD HH:MM:SS) as Date. Just FYI, even though allowed, don't consider it as a good practice to contain space in the column name.

Answer 2

In base R:

my.data <- read.table(text = '

 JulianDay Azimuth   Zenith  Date.YYMMDD Date.HHMMSS
 2455928 174.14066 70.04650 2012-01-01 13:00:00
 2455928 188.80626 70.30747 2012-01-01 14:00:00
 2455928 203.03458 73.12297 2012-01-01 15:00:00
 2455928 216.28061 78.20131 2012-01-01 16:00:00
 2455928 228.35929 85.10759 2012-01-01 17:00:00
 2455929 160.00000 70.04650 2012-01-02 13:00:00
 2455929 188.80626 70.30747 2012-01-02 14:00:00
 2455929 203.03458 73.12297 2012-01-02 15:00:00
 2455929 216.28061 78.20131 2012-01-02 16:00:00
 2455929 228.35929 85.10759 2012-01-02 17:00:00
', header = TRUE)

with(my.data, aggregate(Azimuth ~ JulianDay, FUN =  function(x) c(Min = min(x), Max = max(x))))

One problem with aggregate is that the output is not is a form that is easy to use. It requires a bit of post processing:

my.min.max <- with(my.data, aggregate(my.data$Azimuth, by = list(my.data$JulianDay), 
                   FUN = function(x) c(MIN = min(x), MAX = max(x)) ))

# to convert output of aggregate into a data frame:

my.min.max2 <- do.call(data.frame, my.min.max)

# combine output from aggregate with original data set

colnames(my.min.max2) <- c('JulianDay', 'my.min', 'my.max')

my.data2 <- merge(my.data, my.min.max2, by = 'JulianDay')
my.data2

#   JulianDay  Azimuth   Zenith Date.YYMMDD Date.HHMMSS   my.min   my.max
#1    2455928 174.1407 70.04650  2012-01-01    13:00:00 174.1407 228.3593
#2    2455928 188.8063 70.30747  2012-01-01    14:00:00 174.1407 228.3593
#3    2455928 203.0346 73.12297  2012-01-01    15:00:00 174.1407 228.3593
#4    2455928 216.2806 78.20131  2012-01-01    16:00:00 174.1407 228.3593
#5    2455928 228.3593 85.10759  2012-01-01    17:00:00 174.1407 228.3593
#6    2455929 160.0000 70.04650  2012-01-02    13:00:00 160.0000 228.3593
#7    2455929 188.8063 70.30747  2012-01-02    14:00:00 160.0000 228.3593
#8    2455929 203.0346 73.12297  2012-01-02    15:00:00 160.0000 228.3593
#9    2455929 216.2806 78.20131  2012-01-02    16:00:00 160.0000 228.3593
#10   2455929 228.3593 85.10759  2012-01-02    17:00:00 160.0000 228.3593

You can use by also, but the output from by also requires a bit of post-processing:

by.min.max <- as.data.frame(do.call("rbind", by(my.data$Azimuth, my.data$JulianDay, 
                            FUN =  function(x) c(Min = min(x), Max = max(x)))))

by.min.max <- cbind(JulianDay = rownames(by.min.max), by.min.max)

my.data2 <- merge(my.data, by.min.max, by = 'JulianDay')
my.data2

You can also use tapply:

my.data$Date_Time <- as.POSIXct(paste(my.data$Date.YYMMDD, my.data$Date.HHMMSS), 
                                format = "%Y-%d-%m %H:%M:%S")

ty.min.max <- as.data.frame(do.call("rbind", tapply(my.data$Azimuth, my.data$JulianDay, 
                            FUN =  function(x) c(Min = min(x), Max = max(x)))))

ty.min.max <- cbind(JulianDay = rownames(ty.min.max), ty.min.max)

my.data2 <- merge(my.data, ty.min.max, by = 'JulianDay')
my.data2

You can also use a combination of split and sapply:

sy.min.max <- t(sapply(split(my.data$Azimuth, my.data$JulianDay),
               function(x) c(Min = min(x), Max = max(x)) ))

sy.min.max <- data.frame(JulianDay = rownames(sy.min.max), sy.min.max,
                         stringsAsFactors = FALSE)

my.data2 <- merge(my.data, sy.min.max, by = 'JulianDay')
my.data2

You can also use a combination of split and lapply:

ly.min.max <- lapply(split(my.data$Azimuth, my.data$JulianDay),
                     function(x) c(Min = min(x), Max = max(x)))

ly.min.max <- as.data.frame(do.call("rbind", ly.min.max))

ly.min.max <- cbind(JulianDay = rownames(ly.min.max), ly.min.max)

my.data2 <- merge(my.data, ly.min.max, by = 'JulianDay')
my.data2

You can also use ave, although I have not figured out how to use two functions in one ave statement:

my.min <- ave(my.data$Azimuth, my.data$JulianDay, FUN = min)
my.max <- ave(my.data$Azimuth, my.data$JulianDay, FUN = max)

my.data2 <- data.frame(my.data, my.min, my.max)
my.data2

Answer 3

With dplyr

library(dplyr)
df %>%
    group_by(JulianDay) %>% #if you need `Date` class, use `as.Date(JulianDay)`
    summarise(MaxZenith = max(Zenith), minZenith = min(Zenith))

where 'JulianDay' is the renamed column name for (YYMMDD HH:MM:SS)