491
18.3 Production Workflows for EBVs
The estimation of EBV information products typically involves multiple levels of
data integration, from the collection of raw observations to the production of a final,
consistent information set that provides comparable measurements in space and
time. Data integration procedures need to be customized for almost every EBV,
since they need to accommodate highly diverse biological quantities that are often
specific to a particular EBV. Designing open, consistent, and fully reproducible
workflows is key to support the full operationalization process, from data collection
to publication of an EBV product that is ready to use for multiple science and policy
purposes.
18.3.1 The Need for Open EBV Workflows
Workflows are defined as precise descriptions of data processing from one analytical step to another in a formal language. In recent years a multiplication of biodiversity data availability, novel analytical capabilities, and virtual infrastructures have
laid the foundations for producing better integrated and more detailed information
for measuring biodiversity change (e.g., Jetz et al. 2012; Hansen et al. 2013).
However, the increasing variety of analytical procedures and project-specific
designs also means that analytical standards are difficult to establish (Borregaard
and Hart 2016). Open workflows benefit the preservation of processing steps and
support data interoperability and the automation of biological and environmental
data integration (e.g., via virtual biodiversity e-infrastructures; La Salle et al. 2016).
These workflows require provenance of derived products to be also recorded so others can understand the relationships among data, processing, and results (Michener
and Jones 2012) and thus facilitate product updating as new data and better processing algorithms become available. All these aspects are critical in the EBV framework since the production of EBVs depends on large research collaborations built
on the basis of knowledge transfer and open access to data and production protocols.
At present, fully operational workflows that facilitate the automated and widespread production of EBVs are missing. However, recent efforts have identified
critical steps and bottlenecks for the definition of workflows in support of the production of specific EBVs.
18.3.2 From Data Collection to Biodiversity Models
Generic workflows have been outlined so far for the production of a few speciescentered EBVs, including species distributions, population abundances, and species
traits (Kissling et al. 2018a, b). For example, 11 steps have been identified to build
18 Essential Biodiversity Variables: Integrating In-Situ Observations and Remote…
18.3 Production Workflows for EBVs
The estimation of EBV information products typically involves multiple levels of
data integration, from the collection of raw observations to the production of a final,
consistent information set that provides comparable measurements in space and
time. Data integration procedures need to be customized for almost every EBV,
since they need to accommodate highly diverse biological quantities that are often
specific to a particular EBV. Designing open, consistent, and fully reproducible
workflows is key to support the full operationalization process, from data collection
to publication of an EBV product that is ready to use for multiple science and policy
purposes.
18.3.1 The Need for Open EBV Workflows
Workflows are defined as precise descriptions of data processing from one analytical step to another in a formal language. In recent years a multiplication of biodiversity data availability, novel analytical capabilities, and virtual infrastructures have
laid the foundations for producing better integrated and more detailed information
for measuring biodiversity change (e.g., Jetz et al. 2012; Hansen et al. 2013).
However, the increasing variety of analytical procedures and project-specific
designs also means that analytical standards are difficult to establish (Borregaard
and Hart 2016). Open workflows benefit the preservation of processing steps and
support data interoperability and the automation of biological and environmental
data integration (e.g., via virtual biodiversity e-infrastructures; La Salle et al. 2016).
These workflows require provenance of derived products to be also recorded so others can understand the relationships among data, processing, and results (Michener
and Jones 2012) and thus facilitate product updating as new data and better processing algorithms become available. All these aspects are critical in the EBV framework since the production of EBVs depends on large research collaborations built
on the basis of knowledge transfer and open access to data and production protocols.
At present, fully operational workflows that facilitate the automated and widespread production of EBVs are missing. However, recent efforts have identified
critical steps and bottlenecks for the definition of workflows in support of the production of specific EBVs.
18.3.2 From Data Collection to Biodiversity Models
Generic workflows have been outlined so far for the production of a few speciescentered EBVs, including species distributions, population abundances, and species
traits (Kissling et al. 2018a, b). For example, 11 steps have been identified to build
18 Essential Biodiversity Variables: Integrating In-Situ Observations and Remote…