Traits Use cases.html

Participants

• Mark*, Shawn, Steve, Damian, Benno, Flip

Use Case 1: Predicting species occurrence based on point data (e.g., bird counts) and other biotic & abiotic data. Want to estimate population densities (presence/absence; abundance) of birds (just as an example; others include specimen location information, road-kill sitings, etc.) based on growing database of organismal occurrence data (species/location/date). 

Assist researchers in locating and assembling for further analyses, data sets of covariates (predictor variables) to inform models of occurrence and abundance. Scope expansion-- if one has certain species, what behavioral/life history characteristics are available for these?

CONCRETE: integrate bird point data with MODIS data with meteorological data.  Determine programmatic access to these sources.  (Maybe take lat/lon and semantic map that to Obs model to unify lat/lon anywhere appears, e.g. met data and bird point data).

eBird mapped to ObsOnt.  Weather underground or IRI Data lib mapped to ObsOnt-- keying on lat/lon in both.

phylogenetic commu analysis, trait-commu analysis use cases to be developed?

• Still need to elaborate upon useful linkages with genomics and other ecological/behavioral/trait observational data (nesting location and timing, dietary characteristics), via ontologies.
  
• Summary
• Queries:
  
• 1) what variables are available as covariates for my occurrence data-- is there information about mean daily temperature at lat/lon over xx time span?
  
•      thematic search: weather/climate; soil, land-use, human demographics, hydrology, habitat type
  
•     return number of data sets, can be searched by facets:  sector, realm, institution, author, grain/resolution (in space-time) --> semantics of these can be elaborated and referenced from ontology 
  
•     how determine the catalog of measurements that are available, and the provenance of their values-- modeled surface, nearest gage reading, etc.  E.g. there are many ways to estimate a temperature at some given lat/lon.  Lots of datasets doing it for some given lat/lon.  Which should analyst use?  How would semantics guide in resolving these?
  
• what is the nearest freshwater source (stream/pond/lake) to lat/lon?  what is growth rate of human popu over last XX years at lat/lon?  what is land-use pattern at lat/lon?  what are areas having XX diversity of birds at YY time?  what studies about [nesting, feeding] in bird sp. X have been done in region Y?
  
• Operations/Tasks:
  
• must have an easy way of discovering and acquiring  useful abiotic and biotic data "coverages" to associate with point data; express model outputs as some standard set of coverages to inform other analyses
• enable flexible querying of a variety of geospatial coverage data relative (expressed in common projection with access to " catalog" of well-defined attributes, which reveal capacity to drill-down or roll-up those measurements, and include details as to their provenance-- owner, methodology for collection, etc.) to those point data in order to better understand underlying ecological drivers for those densities; enable extraction of coverage data as single-value supplements to table (e.g. landuse=agric; drill-down to soybean farm)
• must have flexible ways to define 'co-location'-- interest might be in radius of importance of features for breeding birds-- proximity to food, water, shelter.
•  requires extracting coverage data values of a variety of abiotic and biotic data (land use, human census info, meteorological data, with flexible radii of relevance.  E.g. proximity to freshwater-- 1km away from 3 major bodies of water or 100m away from small stream. Temporal aspects of importance as well.
• Need to incorporate specific data sets and associated metadata: NLCD; MODIS, "individual researcher" observational data sets and intepretation of attributes from these
• General Integration Challenges--
  
•     resolve scale issues: everything has to be at same resolution--system must be able to aggregate so that can union commensurate measures of data.
•     measurement scale alignment-- simple example fahrenheit/celsius
•     rebin/rescale grids to integrate on common scale-- semantic specification of operation across data sets?
  
•     Ecological genetics-- need to engage some practitioners  to help flesh this out. 
  
•     Trait-based community analysis-- need practitioners to help define
•     thematic alignment-- attributes referred to using synonyms, similar-to, coarser-- resolve with ontologies (need specific example)
  
•     Tagging and retrieving data based on epiphenomena-- e.g. northern migrational front/pattern;  involves tracking delta/changes-- when are birds in Mississippi River Delta; when are birds reaching Great Lakes; generally indicating trends--** feedback from analysis/models into metadata/data.  Like in genomics commu-- include both highly structured tagging and idiosyncratic taging.
  
• Metrics of completion/success
•     We refine some of above points, and predetermine subset of results.  Challenge expected to automate more comprehensive results (larger number of coverages discovered, selected, rescaled, integrated

SPECIFIC DATA and METADATA for challenge