R help - Sep 2017 - compounding precipitation based on whether falls within a day

Using the small reproducible example below, I'm wondering how best to
complete the following task:

In the small reproducible example below, the 3D array prec has indexes that
correspond to time, x, y (i.e., prec[time, x, y]).  In this case, the time
index is hours since some predefined start time.  I'd like to add up all
the time indexes in 'prec' based on whether or not the corresponding
hours
fall within a given day.  So, at the end of the small example below, there
are two variables that I'm left with, prec_idx (an hourly sequence from
beg_time to end_time) whose length is equal to the first index (the time
index) of the 3D array of precipitation called prec.  That is, I'd like to
get a 3D array called prec*_daily* that has dimension prec*_daily*[21, 3, 4],
where 21 is the number of days and the value in say prec*_daily*[1,x,y] is
equal to prec[1,x,y] + prec[2,x,y] + ... + prec[24,x,y]


ndays <- 21
base_time <- as.character('2001-12-31T23:00:00Z')
hrs_since_base <- 1

# adding an extra second to the end b/c I'm paranoid about the midnight
time stamp not being explicit
beg_time <- as.POSIXct(base_time, format = "%Y-%m-%dT%H:%M:%SZ") +
(hrs_since_base * 60 * 60) + 1

max_hr_since <- 24 * ndays
end_time <- as.POSIXct(base_time, format = "%Y-%m-%dT%H:%M:%SZ") +
(max_hr_since * 60 * 60) + 1

prec_idx <- seq(beg_time, end_time, by='hour')

prec <- array(abs(rnorm((24*ndays) * 3 * 4)) , dim=c(24*ndays, 3, 4))

length(prec_idx)
# 504
dim(prec)
# 504   3   4

# How do I aggregate prec to get daily sums of precipitation based on the
prec_idx array?

	[[alternative HTML version deleted]]

Thanks for the reprex. Wouldn't have bothered without it.

The following is I believe **almost** what you want. It seems a bit clumsy
to me, so others may provide you something neater. But anyway...

## Convert POSIXct vector to dates
## There are 22 different days, not 21
date <- as.Date(prec_idx)

## Sum results by date at each i,j of the last 2 array dimensions
z <- lapply(unique(date),function(d)
   apply(prec[date==d,,],2:3,sum)
   )

## This gives a list with 22 3x4 matrices of sums.
## Convert to 3x4x22 array with

prec_daily <- array(unlist(z),dim=c(3,4,22))

## This is the **almost** part. You can use the aperm() function to reshape
the array if you like. I leave those pleasures to you.

HTH.

Cheers,
Bert



Bert Gunter

"The trouble with having an open mind is that people keep coming along and
sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )

On Wed, Sep 13, 2017 at 9:52 AM, Morway, Eric <emorway at usgs.gov> wrote:
> Using the small reproducible example below, I'm wondering how best to
> complete the following task:
>
> In the small reproducible example below, the 3D array prec has indexes that
> correspond to time, x, y (i.e., prec[time, x, y]).  In this case, the time
> index is hours since some predefined start time.  I'd like to add up
all
> the time indexes in 'prec' based on whether or not the
corresponding hours
> fall within a given day.  So, at the end of the small example below, there
> are two variables that I'm left with, prec_idx (an hourly sequence from
> beg_time to end_time) whose length is equal to the first index (the time
> index) of the 3D array of precipitation called prec.  That is, I'd like
to
> get a 3D array called prec*_daily* that has dimension prec*_daily*[21, 3,
> 4],
> where 21 is the number of days and the value in say prec*_daily*[1,x,y] is
> equal to prec[1,x,y] + prec[2,x,y] + ... + prec[24,x,y]
>
>
> ndays <- 21
> base_time <- as.character('2001-12-31T23:00:00Z')
> hrs_since_base <- 1
>
> # adding an extra second to the end b/c I'm paranoid about the midnight
> time stamp not being explicit
> beg_time <- as.POSIXct(base_time, format =
"%Y-%m-%dT%H:%M:%SZ") +
> (hrs_since_base * 60 * 60) + 1
>
> max_hr_since <- 24 * ndays
> end_time <- as.POSIXct(base_time, format =
"%Y-%m-%dT%H:%M:%SZ") +
> (max_hr_since * 60 * 60) + 1
>
> prec_idx <- seq(beg_time, end_time, by='hour')
>
> prec <- array(abs(rnorm((24*ndays) * 3 * 4)) , dim=c(24*ndays, 3, 4))
>
> length(prec_idx)
> # 504
> dim(prec)
> # 504   3   4
>
> # How do I aggregate prec to get daily sums of precipitation based on the
> prec_idx array?
>
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/
> posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
	[[alternative HTML version deleted]]

Hi Eric,
Bert's solution is very elegant. His final comment prompted me to check out
the aperm() function which I have never used.
The final step to complete his response is

prec_daily2 <- aperm(prec_daily, c(3,1,2))

Regards


On Wed, Sep 13, 2017 at 9:06 PM, Bert Gunter <bgunter.4567 at gmail.com>
wrote:
> Thanks for the reprex. Wouldn't have bothered without it.
>
> The following is I believe **almost** what you want. It seems a bit clumsy
> to me, so others may provide you something neater. But anyway...
>
> ## Convert POSIXct vector to dates
> ## There are 22 different days, not 21
> date <- as.Date(prec_idx)
>
> ## Sum results by date at each i,j of the last 2 array dimensions
> z <- lapply(unique(date),function(d)
>    apply(prec[date==d,,],2:3,sum)
>    )
>
> ## This gives a list with 22 3x4 matrices of sums.
> ## Convert to 3x4x22 array with
>
> prec_daily <- array(unlist(z),dim=c(3,4,22))
>
> ## This is the **almost** part. You can use the aperm() function to reshape
> the array if you like. I leave those pleasures to you.
>
> HTH.
>
> Cheers,
> Bert
>
>
>
> Bert Gunter
>
> "The trouble with having an open mind is that people keep coming along
and
> sticking things into it."
> -- Opus (aka Berkeley Breathed in his "Bloom County" comic strip
)
>
> On Wed, Sep 13, 2017 at 9:52 AM, Morway, Eric <emorway at usgs.gov>
wrote:
>
> > Using the small reproducible example below, I'm wondering how best
to
> > complete the following task:
> >
> > In the small reproducible example below, the 3D array prec has indexes
> that
> > correspond to time, x, y (i.e., prec[time, x, y]).  In this case, the
> time
> > index is hours since some predefined start time.  I'd like to add
up all
> > the time indexes in 'prec' based on whether or not the
corresponding
> hours
> > fall within a given day.  So, at the end of the small example below,
> there
> > are two variables that I'm left with, prec_idx (an hourly sequence
from
> > beg_time to end_time) whose length is equal to the first index (the
time
> > index) of the 3D array of precipitation called prec.  That is, I'd
like
> to
> > get a 3D array called prec*_daily* that has dimension prec*_daily*[21,
3,
> > 4],
> > where 21 is the number of days and the value in say
prec*_daily*[1,x,y]
> is
> > equal to prec[1,x,y] + prec[2,x,y] + ... + prec[24,x,y]
> >
> >
> > ndays <- 21
> > base_time <- as.character('2001-12-31T23:00:00Z')
> > hrs_since_base <- 1
> >
> > # adding an extra second to the end b/c I'm paranoid about the
midnight
> > time stamp not being explicit
> > beg_time <- as.POSIXct(base_time, format =
"%Y-%m-%dT%H:%M:%SZ") +
> > (hrs_since_base * 60 * 60) + 1
> >
> > max_hr_since <- 24 * ndays
> > end_time <- as.POSIXct(base_time, format =
"%Y-%m-%dT%H:%M:%SZ") +
> > (max_hr_since * 60 * 60) + 1
> >
> > prec_idx <- seq(beg_time, end_time, by='hour')
> >
> > prec <- array(abs(rnorm((24*ndays) * 3 * 4)) , dim=c(24*ndays, 3,
4))
> >
> > length(prec_idx)
> > # 504
> > dim(prec)
> > # 504   3   4
> >
> > # How do I aggregate prec to get daily sums of precipitation based on
the
> > prec_idx array?
> >
> >         [[alternative HTML version deleted]]
> >
> > ______________________________________________
> > R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> > https://stat.ethz.ch/mailman/listinfo/r-help
> > PLEASE do read the posting guide http://www.R-project.org/
> > posting-guide.html
> > and provide commented, minimal, self-contained, reproducible code.
> >
>
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/
> posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
	[[alternative HTML version deleted]]