Epic fail in MCZ Darwin Core with "unknown" dates

15909
Reporter: rdmpage
Type: Bug
Summary: Epic fail in MCZ Darwin Core with "unknown" dates
Priority: Major
Status: Open
Created: 2014-06-07 17:41:13.865
Updated: 2015-03-02 15:04:04.742
        
Description: This looks like it's not a GBIF bug as such, but there are GBIF-related issues.

The MCZ dataset http://www.gbif.org/dataset/4bfac3ea-8763-4f4b-a71a-76a6f5f243d3 has a lot of dates that are "1700-01-01". Despite them being recorded as this in the MCZDarwin Core Archive, many (all?) of these are actually "unknown". For example, http://mczbase.mcz.harvard.edu/guid/MCZ:IZ:DIPL-4985 shows the following:

Collecting Date:	[date unknown] (1700-01-01 - 2100-01-01)

(this is http://www.gbif.org/occurrence/699109039 ).

In the DWCA file they supply this becomes "1700-01-01", which GBIF interprets literally. However the "verbatimEventDate" field states "[date unknown]",

So, the MCZ has truncated date export, and GBIFG has ignored the contradiction between having a precise collection date and the value of verbatimEventDate.

This needs to be fixed. Ideally MCZ would output the data correctly, and GBIF would be able to handle uncertainty as a range.

]]>
    

Author: kbraak@gbif.org
Comment: A reminder [~omeyn@gbif.org] to link this issue to the date interpretation issue you're working on. Thanks
Created: 2014-06-11 17:07:27.413
Updated: 2014-06-11 17:07:27.413

Author: rdmpage
Comment: For my own interest, is the related issue POR-2120 ?
Created: 2014-06-11 19:26:10.818
Updated: 2014-06-11 19:26:24.738