PCICt: PCICt, a POSIXct work-alike for 360- and 365-day calendars

In PCICt: Implementation of POSIXct Work-Alike for 365 and 360 Day Calendars

PCICt, a POSIXct work-alike for 360- and 365-day calendars

Description

This package implements a work-alike to R's POSIXct class which implements 360- and 365-day calendars in addition to the gregorian calendar.

Details

Many global climate models (GCMs) and regional climate models (RCMs) are run using an idealized and simplified calendar which only includes 365 days or 360 days per year. When trying to do seasonal or monthly analysis on a set of models which use different calendar types, analyses may not be comparable unless one can normalize the calendars and the times which are represented therin. Thing get even more difficult when trying to compare model output with observations data which is located in a particular time on the Gregorian calendar.

The PCICt package attempts to solve this problem by creating a new time type, PCICt, which inherits from the POSIXt type. All of the functionality provided by POSIXt is also provided by PCICt, however PCICt does the work of normalizing the calendars and making points in time on seperate calendars cross comparable.

365-day calendars are implemented using a 365-day non-leap year from a Gregorian calendar.

360-day calendars are not implemented as 12 equal months of 30 days. They are implemented as 12 months of the following lengths, in days, with the first month being January: 30, 28, 31, 30, 30, 30, 30, 31, 30, 30, 30, 30. This was a decision to ease implementation.

To map a 365_day calendar to Gregorian, PCICt simply drops February 29 from leap years. To map a 360_day calendar to Gregorian, PCICt attempts to remap the days such that the five lost days are distributed as equally as possible across the seasons (winter loses two days while, spring/summer/fall each lose one).

There are a few problems with this implementation. As noted above, 360-day calendars do not use equal months, which may cause problems in certain situations.

Another problem originates from within R itself. Many functions in R strip attributes from data. If this happens to a PCICt object, it cannot be coerced back into a PCICt, as it is lacking the calendar attribute. This causes problems with several internal functions, like mean. PCICt includes a wrapper for mean. You will probably run into these problems. When you do, please use the wrapper for mean as a template for your wrapper.

This package may be modified substantially in future to solve these problems.

References

http://www.pacificclimate.org

See Also

as.PCICt

PCICt documentation built on Feb. 16, 2023, 7:43 p.m.