How to handle sieve fractions

[Daphnisd]

A benthic EMODnet dataset. A sample was treated using 2 sieves: 0.5 mm and 1mm which resulted in 2 fractions. So I proposed the following structure:

             eventID   parentEventID
--------------------   -------------
             Sample1
Sample1:Sieve0.5-1mm         Sample1
   Sample1:Sieve>1mm         Sample1

They will provide both abundances (counts / m2) as original counts.

However: The above structure is conceptually different from events seperated in subsamples. If you want to calculate abundance of a species in a samples based on the subsamples; you need to average the abundances. If you have abundances of different Sieve fractions you will need to sum the abundances.

I see 3 options to solve the issue:

1) We need a way to differentiate between these 2 different types. Can this done based on the field "type"? E.G. using type "Sieve Fraction" or "Sieze Fraction"? Or another way in the data table to differentiate?

2) We could in this case link the abundance to the sample and sum the abundances and omit the detail of the sieze fractions

So structure is the following and abundances summed

eventID  parentEventID
-------  -------------
Sample1

3) We could link all data to same sample1 but list 2 different occurrences. Occurrence 1 has a MoF size class = 0.5-1 mm, occurrence2 has a size class > 1mm. This latter option has the benefit that if you want the total abundance of a species in a sample you allready need to sum the abundances for all sexes and life stages.

So structure is the following

Event

eventID  parentEventID
-------  -------------
Sample1

MoF

occurrenceID              type  value  unit
------------  ----------------  -----  ----
 Occurrence1         abundance      5  #/m2
 Occurrence1  Lower size class    0.5    mm
 Occurrence1  upper size class      1    mm
 Occurrence2         abundance     10  #/m2
 Occurrence2  Lower size class      1    mm

I originally thought option 1 was the solution, but maybe option 3 is better? Or would OBIS support both 1 and 3?

[Daphnisd]

Actually: this situation is similar in meiofauna analyses, where one core is divided in different slices covering different depths is this core. These should be describes as different events, and if you would like to have the total abundance per square meter a user will also need to sum these abundances.

So for the type vocabulary we have 2 new types: slice and sizeFraction

As a guideline; I guess OBIS can say that both option 1 and option 3 are accepted, but option 1 is preferred as this best describes the situation?

Any thoughts?

[pieterprovoost]

It seems to me that option 3 is option 1 but without bothering to create subevents for the size classes. In fact when you "flatten" the option 1 data to occurrences + facts (which is likely what we will do in OBIS), you obtain data in option 3 format.