I often have data files from different sources that I want to join by time, but as they come from different systems their time stamps are just slightly different. We solved this by using JMPs KDTable object which allows for crazy fast nearest neighbor searching. This "fuzzymerge" will find the closest row in time in a second table for each row in the main table and join those rows into the resulting single table.
I found a few other examples of this in my search on the community, but none with a GUI to make it accessible to non-scripters.
To use this, first run the 'fn_fuzzymerge_generic' script to load the custom function into JMP. Then run the fuzzymerge_generic_GUI script with your data tables open. You might want to take a look at a distribution of the 'Fuzzy Merge delta' column which is the difference between the time columns of the two tables for the rows that were joined. Large values indicate that there is no data "close by" that row in the main table.
Note that this will find the closest row *in any direction*, forward or backward in time.
Hope this helps someone!
Kristo