I solved this with a client application. The technique was to create a color cylindrical map of the world, each time zone has a unique color. Lat-Lons are converted to image coordinates, and the color of the coordinate is read, and then cross-linked to the time zone of that color.
This short explanation is not exactly what I did, but he moved on to the idea. I actually filled out a couple of dictionaries and looked for them. The first execution required a 2M resource file to fill out (after I processed my card and turned it into binary data). The theoretical maximum error (for latitudes near the equator) should be +/- about 15 miles. Unfortunately, my initial map accuracy was more like +/- 100 miles.
So I'm redoing the project. I have been doing this for several days, creating both a more accurate and a higher map for a crunch. A couple more days, and this needs to be done. The resource file will be about 20 M, unless I select only the resources of the βstupid areaβ and mathematically calculate 90% of the world that can be calculated directly (most of the world can be mathematically derived from longitude). I'm not sure that many will make sure that the resource file is 20 M, but some could. In any case, if there is any interest, I will try to publish the code necessary to run it, and the resource file is one of the common code sites. If there seems to be no interest, I will not worry.
Just to re-iterate the code needed in your application, there are only a few lines, but the resource file is large (without compression, my new one works at 22 MB). It is also fast (the first performance was 100 M / s). This requires a file download, and it takes a little time. The 2M version did not have a noticeable delay, but the 22M may (not yet).
shooky
source share