Convert from minutes since origin to date time for normal people ("yyyy-mm-dd HH:MM:SS")?

I'm working with a netcdf file which has its time units described as: "minutes since 1850-01-16T14:06:00"

I'd like to have it in a more familiar date format for easier date matching and processing. The origin is given which is helpful, but I can't determine if I am using the wrong parsing formats or if the dates truly are not in line with the expectations I had.

I checked the help for strptime() and as.POSIXct() and haven't been able to figure out what format to use to parse this bizarre time structure.

# Example of the origin format and first 8 values
origin <- "1850-01-16T14:06:00"
dates_as_times <- c(52559874, 52602354, 52644834, 52688754, 
                    52732674, 52776594, 52820514, 52865154)

Solution

There are some packages for converting the netcdf files

library(RNetCDF)
library(lubridate)
out <- utcal.nc("minutes since 1850-01-16T14:06:00", dates_as_times)
out
#     year month day hour minute second
#[1,] 1949    12  23   12      0      0
#[2,] 1950     1  22    0      0      0
#[3,] 1950     2  20   12      0      0
#[4,] 1950     3  23    0      0      0
#[5,] 1950     4  22   12      0      0
#[6,] 1950     5  23    0      0      0
#[7,] 1950     6  22   12      0      0
#[8,] 1950     7  23   12      0      0

From the matrix, it can be easily converted to Datetime object with ISOdatetime

with(as.data.frame(out), ISOdatetime(year, month, day,
        hour, minute, second, tz = 'GMT'))
#[1] "1949-12-23 12:00:00 GMT" "1950-01-22 00:00:00 GMT" "1950-02-20 12:00:00 GMT" "1950-03-23 00:00:00 GMT"
#[5] "1950-04-22 12:00:00 GMT" "1950-05-23 00:00:00 GMT" "1950-06-22 12:00:00 GMT" "1950-07-23 12:00:00 GMT"

Or another option is

library(ncdf.tools)
convertDateNcdf2R(dates_as_times, origin = ymd_hms(origin))

Or we could multiply with 60

as_datetime(dates_as_times * 60, origin = ymd_hms(origin))