Steve
Looking at the web site and thinking about definitions and terms - I am not entirely happy with the use of the term 'Negative data'. In most of the areas I have been involved, the term 'Absence data' is more common and I think is more self explanatory (even though it is really implied absence rather than true absence in most cases).
Cheers
Arthur D. Chapman
From Steve Kelling stk2@cornell.edu on 9 Feb 2006:
Hello, I really like the improvements made on the definition, and I might suggest that I put it up on the TDWG-obs website http://www.avianknowledge.net/tdwg. I think we still need to define occurrence, and I think that Arthur Chapman's "species occurrence data" gets us most of the way there. My sense is that the elaborations need to be expanded a bit more. For example, data collection event. I suggest that we include in the definition some information about time, minimum required data, and so forth. These definitions can be longer and more detailed than the observation definition. So taking data collection event: An event, during or after which at least the minimum required data were recorded.
Can we identify the minimum required data? For us it is: who the collector is where the information was collected (and should refer to occurrence information and all of the discussion with it) when the data and time what was observed (and how many) effort distance covered, time spent etc. I'm sure there are other data for the minimum required data and this is worth discussion.
We should also discuss the event. For example, one of ours and Bird Studies Canada projects (Project FeederWatch) the event can last 2 days.
We also need to think about whether we want to include protocol. In our definitions no protocol (incidental observation) is also a protocol.
Anyway, thanks for the improvements on the definition of observation. I (like Lynn) look forward to discussing some of these other issues.
Steve
At 05:03 PM 2/8/2006 -0500, you wrote:
With only comments from Bob & Arthur (thanks!), the latest version of observation definitions is as follows:
"An observation characterizes the evidence for the presence or absence of an organism or set of organisms through a data collection event at a location. Observations are not necessarily independent and could be linked via characteristics such as time, place, protocol, and co-occurring organisms."
With the following draft elaborations for various terms (many thanks to Bob Peet for providing most of these draft definitions):
- occurrence
"Occurrence" has been changed to "evidence for the presence or
absence".
The key idea is that the organism or set of organisms was either detected or not. We also need to provide an opportunity for the recorder to note the certainty.
As an aside, recall we need to support minimalist protocols (e.g. "organism/community (not)seen in field", "organism heard in field", "scat seen in field", "tracks seen in field", "museum collection".)
- data collection event
An event, during or after which at least the minimum required data were recorded.
- location
Ideally, at least geocoordinates plus an accuracy term. Since there is
a
considerable amount of historical / legacy data that does not presently have a georeference yet has valuable information that should be
included
in observation databases and shared, we cannot at this time require
data
to be in a GIS format.
I (Lynn) suggest: (a) Location information be required, preferably geocoordinates and mapping precision, but if not available then a text description and the finest level of geolocation using the Darwin Core attributes.
(b) Location data include the representation of observations as point, line, or polygon data (with the necessary spatial metadata).
- entity
Dropped from the definition of observation.
- could be linked
Can have a pointer or pointers to other observations, thereby creating aggregate observations. Note that commonality of date, time, place, etc. is not sufficient in that the none of the observation authors explicitly made the connection
Please share your comments / thoughts on all of the above definitions with this email list.
If people are generally comfortable with the above as working defintions, then I'd like to propose that we move into the fun part of identifying attributes to be developed into a schema.
=== message truncated ===