Converts a mixed Julian/Gregorian date to a UT-referenced date (deprecated; use cd_calendar).
function ut_calendar ( time : numeric, option  : integer )
A multi-dimensional array of time values in a mixed Julian/Gregorian, 360, 365, or no leap calendaring system (the last three options were added in version 5.1.0).
May additionally have a "calendar" attribute with one of the following values:
No other calendaring systems are recognized by this function, including:
command_year n kyr B.P.option
A scalar integer indicating the format of the output. See description below.
The format of the output is dependent upon the value of option. In all cases the first dimension is the same size as time.
- option = 0
The array returned will be of type float and dimensioned dimsizes(time) x 6:
utc_date(:,0) --> years
utc_date(:,1) --> months
utc_date(:,2) --> days
utc_date(:,3) --> hours
utc_date(:,4) --> minutes
utc_date(:,5) --> seconds
- option = -1 or 1
The values returned will be in the format YYYYMM and will be type double for option = 1, and type integer for option = -1.
Note that for option = 1, even though the return value is double, the value will be the same as if an integer had been returned (no fraction is returned, so days, hours, minutes, and seconds are basically truncated).
- option = -2 or 2
The values returned will be in the format YYYYMMDD and will be type double for option = 2, and type integer for option = -2.
Note that for option = 2, even though the return value is double, the value will be the same as if an integer had been returned (no fraction is returned, so hours, minutes, and seconds are basically truncated).
- option = -3 or 3
The values returned will be in the format YYYYMMDDHH and will be type double for option = 3, and type integer for option = -3. Note that this option can produce some big numbers. If your year values go higher than 2147, then then you should use option 3.
Note that for option = 3, even though the return value is double, the value will be the same as if an integer had been returned (no fraction is returned, so minutes, and seconds are basically truncated).
- option = 4
The values returned will be in the format YYYY.fraction_of_year and will be type double. fraction_of_year is the total number of seconds in the current day of the year, divided by the total number of seconds in that year.
- option = -5
Available in version 5.1.0 and later.
The same as option=0, except the values returned will be integers. This means that the seconds values may be truncated, as they are floating point values.
Unidata no longer supports the internal code that ut_calendar is based on. We therefore strongly recommend that you use the cd_calendar function instead. Important note: ut_calendar and cd_calendar treat "year 0" differently. See the caveats below.
Converts a mixed Julian/Gregorian date to a UT-referenced date using the UDUNITS-2 library.
As of version 5.1.0, other calendars are also recognized, if time contains a "calendar" attribute set to one of these values:
- "360", "360_day"
- "365", "365_day"
Thanks to David W. Pierce, the developer of ncview, for providing these calendar additions.
To quote the UDUNITS-2 man page:
The udunits(3) package uses a mixed Gregorian/Julian calendar system. Dates prior to 1582-10-15 are assumed to use the Julian calendar, which was introduced by Julius Caesar in 46 BCE and is based on a year that is exactly 365.25 days long. Dates on and after 1582-10-15 are assumed to use the Gregorian calendar, which was introduced on that date and is based on a year that is exactly 365.2425 days long. (A year is actually approximately 365.242198781 days long.) Seemingly strange behavior of the udunits(3) package can result if a user-given time interval includes the changeover date. For example, ut_calendar and ut_inv_calendar can be used to show that 1582-10-15 *preceded* 1582-10-14 by 9 days.
Caveats of UDUNITS-2:
- Year 0 is treated as year 1, because year 0 does not exist in the
real world calendar.
- The length of a month is fixed at 1/12 of a tropical year or
2629743.831225 seconds. This means if you have a units of something
like "months since 1870-1-1", then at time = 0 you will get:
year = 1870 month = 1 day = 1 hour = 0 second = 0However, at time = 1, you will get:
year = 1870 month = 1 day = 31 hour = 10 second = 3.83122
- Udunits doesn't like uppercase values in the "units" string. For example, "HOURS since 1-1-1 00:00:0.0" is not a valid string, but "hours since 1-1-1 00:00:0.0" is.
- ut_calendar is based on an external software
which has phased out the support for these types of calendaring
routines. Caution should be used when using this routine with special
calendars like the above, or negative values. If you have problems,
then check out the user-contributed
- As of NCL version 5.2.0, this code is being linked against
UDUNITS-2, which has a dependency on some xml files that come installed
with NCL. There's a bug in 5.2.0 that if you are setting the
environment variable UDUNITS_PATH, you may see the error message:
warning:ut_calendar: Invalid specification string. Missing values will be returned.There are two possible work-arounds:
- Set the UDUNITS2_XML_PATH environment variable:
setenv UDUNITS2_XML_PATH $NCARG_ROOT/lib/ncarg/udunits/udunits2.xml or export UDUNITS2_XML_PATH=$NCARG_ROOT/lib/ncarg/udunits/udunits2.xml
- Unset the UDUNITS_PATH environment variable before you run NCL.
Assume five time values that represent "hours since 1-1-1 00:00:0.0". Convert these values to UTC time in the format "hhZ dd mmm yyyy", where "mmm" represents a month abbreviation (and not a numerical month). sprinti is used to create the desired string:
begin ; ; Array to hold month abbreviations. Don't store anything in index ; '0' (i.e. let index 1=Jan, 2=Feb, ..., index 12=Dec). ; month_abbr = (/"","Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep", \ "Oct","Nov","Dec"/) ; ; Time values and units. ; time = (/17522904, 17522928, 17522952, 17522976, 17523000/) time@units = "hours since 1-1-1 00:00:0.0" ; ; Convert to UTC time. ; utc_date = ut_calendar(time, 0) ; ; Store return information into more meaningful variables. ; year = tointeger(utc_date(:,0)) ; Convert to integer for month = tointeger(utc_date(:,1)) ; use sprinti day = tointeger(utc_date(:,2)) hour = tointeger(utc_date(:,3)) minute = tointeger(utc_date(:,4)) second = utc_date(:,5) ; ; Write out strings in the format "hhZ dd mmm yyyy". ; date_str = sprinti("%0.2iZ ", hour) + sprinti("%0.2i ", day) + \ month_abbr(month) + " " + sprinti("%0.4i", year) print(date_str) end
The above script will produce the following output:
Variable: date_str Type: string Total Size: 20 bytes 5 values Number of Dimensions: 1 Dimensions and sizes:  Coordinates: (0) 00Z 01 Jan 2000 (1) 00Z 02 Jan 2000 (2) 00Z 03 Jan 2000 (3) 00Z 04 Jan 2000 (4) 00Z 05 Jan 2000Example 2
Using the same time values as above, here's what the various options return:
dym = ut_calendar(time, 1) ; Double array of length 5 with all ; values equal to 200001. iym = ut_calendar(time, -1) ; Same, only type integer dymd = ut_calendar(time, 2) ; (/20000101,20000102,20000103, iymd = ut_calendar(time, -2) ; 20000104,20000105/) dymdh = ut_calendar(time, 3) ; (/2000010100,2000010200,2000010300, iymdh = ut_calendar(time, -3) ; 2000010400,2000010500/) yearfrac = ut_calendar(time, 4) ; (/2000,2000.002732240437, ; 2000.005464480874, ; 2000.008196721311, ; 2000.010928961749/)Example 3
Use ut_calendar and day_of_year to create a new time variable with units of day_of_year.fraction_of_day. day_of_year requires integer arguments. Use tointeger to convert. To make the results as precise as possible the calculation is done in double precision.
; ; Time values and units. ; time = (/3356, 3356.083, 3356.333, 3356.917, 3357.042, 3358.208/) time@units = "days since 1995-01-01 00:00:0.0" utc_date = ut_calendar(time, 0) dyear = day_of_year(tointeger(utc_date(:,0)) \ ,tointeger(utc_date(:,1)) \ ,tointeger(utc_date(:,2)) )*1.d0 ; make double dyear = dyear + \ (date(:,3)*3600.d0 + date(:,4)*60.d0 + date(:,5)*1.d0)/86400.d0 ; (/70, 70.08299768518519, ; 70.33299768518519, ; 70.91699074074074, ; 71.04199074074074, ; 72.20799768518519/)Example 4
As of version 5.1.0, you can use option = -5 (instead of option = 0) to return integers:
; ; Time values and units. ; time = (/3356, 3356.083, 3356.333, 3356.917, 3357.042, 3358.208/) time@units = "days since 1995-01-01 00:00:0.0" utc_date = ut_calendar(time, -5) dyear = day_of_year(utc_date(:,0),utc_date(:,1),utc_date(:,2))
CAVEAT: The one 'subtle' difference between CAM 'time/date' and NCL's ut_calendar and ut_inv_calendar function occurs when the time units are ".... since 0000..."
NCL uses Unidata's udunits and the updated ncview calendar software. Technically, in the 'time' world, there is no such thing as year 0000. The difference is illustrated by the following:
ncl 0> f = addfile ("camdev07_cam3_6_15_gust_up00.cam2.h1.0000-09-01-00000.nc","r") ncl 1> time = f->time ; days since 0000-09-01 00:00:00 ncl 2> printVarSummary(time) Variable: time Type: double [snip] Dimensions and sizes: [time | 30] Coordinates: time: [ 0.. 29] Number Of Attributes: 4 long_name : time units : days since 0000-09-01 00:00:00 <*** note 0000 calendar : noleap bounds : time_bnds ncl 3> date = f->date ; yyyymmdd [CAM] ncl 4> DATE = ut_calendar(time, -2) ; YYYYMMDD [NCL] ncl 5> DATE1 = DATE - 10000 ; subtract one YYYY ncl 6> print(date+" "+DATE+" "+DATE1) CAM NCL NCL-one_YYYY (0) 901 10901 901 (1) 902 10902 902 (2) 903 10903 903 (3) 904 10904 904 (4) 905 10905 905 (5) 906 10906 906 (6) 907 10907 907 (7) 908 10908 908 (8) 909 10909 909 (9) 910 10910 910 [snip]As noted in the documentation ...
Caveats of Unidata's Udunits: * Year 0 is treated as year 1, because year 0 does not exist in the real world calendar.Punch line: if you use the CAM file's date variable you can readily break down to other units. NCL's ut_calendar function is not needed.
date = f->date ; CAM date (yyyymmdd) year = date/10000 mmdd = date-year*10000 month = mmdd/100 day = mmdd-month*100Again, the difference between CAM and NCL calendars only occurs when the units are "xxxx since 0000".
- Set the UDUNITS2_XML_PATH environment variable: