[Babase] rain gauge rounding

Niki Learn nlearn at princeton.edu
Wed Jul 29 13:19:28 EDT 2009


-----Original Message-----
From: babase-bounces at eeblistserv.Princeton.EDU
[mailto:babase-bounces at eeblistserv.Princeton.EDU] On Behalf Of Karl O. Pinc
Sent: Wednesday, July 29, 2009 12:39 PM
To: The Baboon Database Project
Subject: Re: [Babase] rain gauge rounding

On 07/29/2009 11:10:01 AM, Niki Learn wrote:
> For those out of the loop, we have been wondering why the rain gauge
> data
> (from the handwritten min-max data sheets) is recorded in integers in
> babase
> with no decimal places since this results in a loss of precision and,
> in
> some years, increases annual rainfall by as much as 5mm.
> 

As long as we're revisiting this issue again....

As far as the database goes the significant issue is validation,
particularly with respect to values converted from english
to metric.  If the value of a measurement taken in 100's
of an inch is 3.1415976 when converted to millimeters,
accurate to 1/100th of an inch, but we want to keep
data read in metric with a precision of a 10th of
a mm then we need different rules regarding how many
decimal places depending on weather the reading was
in english or metric units.  This is why the original
design had an indication of the units of measurement
(Inches or mm, or in the case of temperature, F or C)
and the data was to be stored in the original units.
Automatic conversion could then be done to standard
units.

If you're not quite so interested in maximal precision
you can round everything out to a given number of
decimal places after conversion to metric and not
bother storing the original measurement.  The only
question then becomes how many decimal places to keep.


Karl <kop at meme.com>
Free Software:  "You don't pay back, you pay forward."
                 -- Robert A. Heinlein


Niki writes:

Right, well, it seems to me that the closest we can practically get to
1/10th of an inch is still 1/10 of a mm, precisionwise.  Also, from what I
just read about this discussion concerning the temperature conversion from F
to C, the data were not available in F electronically so the whole idea of
entering the older data in F and having babase convert it was set aside due
to the extra work it would require to re-enter the data in F.  From what I
can tell this is also true of the rain measurements in inches - they are
only available electronically in mm.  Obviously it does not make sense to
use two decimals after converting to mm because this introduces an extra
significant figure that implies a degree of accuracy that is clearly
impossible with these rain gauges and a multitude of observers; therefore,
it makes practical sense to round the data converted from inches to the same
number of decimals as the data originally in mm, i.e., to one decimal place.

>From what I can tell Tabby uploaded a file that contains the decimals, which
means that babase then rounded them - correct?  So the data is stored with
the decimals and if we want to change to one decimal place instead of
integers, we can just change the number of decimals to which babase rounds,
right?

Niki



More information about the Babase mailing list