Dear all
Julian Hill and I have further discussed what metadata he needs to describe
night marine air temperature anomalies.
Requirements: (1) To indicate that a statistic for a quantity (specifically
temperature) is evaluated from night-time values only. There is no need at
the moment for a standardised method to record a precise definition of night-
time. (2) To indicate that a quantity is the anomaly from the climatological
value of the statistic. Anomalies are computed in various ways. Though the
general idea is that it is a difference, various treatments are possible. At
present there is no need to record the precise method, which is regarded as
a property of the dataset, so it will be sufficient to identify the source of
the data and/or give references using attributes already defined in CF.
Proposal:
(1) In cell_methods, allow an entry for a time-axis to take the form
"NAME: METHOD where night_time" in which, as usual, NAME identifies the axis
and METHOD is the statistical method e.g. "mean". This use of "where" is
consistent with my other recent posting on cell_methods
http://www.cgd.ucar.edu/pipermail/cf-metadata/2006/001449.html
If "night_time" needs further definition, this can be done in a non-
standardised comment, which a generic application would not interpret e.g.
cell_methods="time: mean where night_time (local sunset + 1hour to local sunrise + 1hour)"
(2) Define a new standard_name qualifier of anomaly_from_climatology, to
indicate that the data values indicate the departure from the climatological
values (representative over some years) of the same statistic and quantity e.g.
standard_name="air_temperature anomaly_from_climatology".
Comments welcome.
Cheers
Jonathan
Received on Mon Jan 15 2007 - 01:17:26 GMT
