TMG-L Archives

Archiver > TMG > 2011-02 > 1297735808

From: "Darrell A. Martin" <>
Subject: Re: [TMG] Latitude and Longitude Conversion
Date: Mon, 14 Feb 2011 20:10:08 -0600

On 2/14/2011 2:42 PM, John Cardinal wrote:
> Darrell A. Martin wrote:
>> I believe the real issue is that the current TMG conversion
>> algorithm is erroneous in ways that are not explained by (proper)
>> rounding. A human can easily convert to and from
>> degrees/minutes/seconds.decimal and degrees.decimal with only a
>> rounding error in the last place of the less precise format, using a
>> simple calculator. Standardized mathematical rules for acceptable
>> precision in this kind of conversion have been around for centuries.
>
> Darrell,
>
> No, that's not true. TMG can do math just fine. The problem is that TMG's
> internal format uses *integer* values for degrees, minutes, and seconds, and
> so it's precision is limited. It's not a math problem, it's a limitation
> based on the format that was chosen to store lat/long values.
>
>> The move toward decimal representation is interesting, of course,
>> but irrelevant to the conversion issue.
>
> No, it's not. Decimal degrees format uses fractional degrees, and only a
> small subset of those values will map exactly to integer values of
> degrees-minutes-seconds. All other values will lose precision during the
> conversion process, but not because of any bug or "conversion issue".
>
> I am not sure why this topic has generated so much conversation. The whole
> thing is pretty simple: if you have a lat/long value in decimal degrees
> format, and you want to record the precise value in TMG, use the tilde
> notation that I have mentioned here and described in the Second Site help
> pages. If you allow TMG to convert the value to its internal format, you
> will lose precision, and for some locations, that is not desirable.
>
> John

John:

Somewhere in the conversation somebody probably mentioned what the
internal format for TMG's storage of lat/long data is, and I missed it.
Regardless, in my opinion the internal storage format is not as
important as the displayed format(s). Users do not typically see the
internal format for lat/long any more than we see the internal format
for dates. The display format(s) are what we work with, and so are the
ones that are important to us. TMG's display formats for lat/long
provide an *apparent* level of mathematical precision which is not
appropriate, given the limits of the way the data is stored. (I am
depending on reported behavior for this statement; I have not tested it.
However, you have not said anything that contradicts it.)

I think the reason this thread has created so much interest is the
effect the handheld GPS has had on locating old cemeteries, abandoned
homesites, etc. It is a potentially exciting tool. Just like DNA
genealogy, some people don't want to think too hard about the limits
that the tool currently has. They want to plug in the coordinates, walk
until their GPS says "we are there" and be standing on the grave.

Lastly, a quibble. I did not say decimal degree representation was
*itself* irrelevant. It is obviously directly relevant. I said the
current *trend* toward decimal representation is irrelevant, which is a
horse of a different color. Bringing the trend into the picture may be
taken to imply that the nature of the problem is changing; or, that if
we wait long enough maybe the issue will just go away ... neither of
which is true, IMO.

Darrell