606
withdrew at a national level, is one important example. Other actors such as basin
commissions, watershed associations, energy cooperatives, and third- party certification programs can fill the gap with forward-looking initiatives that target specific
problems. For these efforts to bear fruit, policymakers must be willing to listen to the
scientific community, and scientists must be able to communicate the results of their
work in language accessible to policymakers and the general public.
21.4.3 Data Fusion Case Study
The FEWSION™ project (https://fewsion.us) is a data fusion effort funded in 2016
by the National Science Foundation’s interdisciplinary INFEWS program
(Innovations at the Nexus of Food Energy and Water Systems). FEWSION brings
together a large number of academic and government data sources to describe the
commodity flows in the US FEW system in a single seamless dataset. This data
enables place-based researchers and those studying a single component of the system to place their work within the broader perspective of the entire FEW system.
This data fusion requires expertise in a large number of distinct datasets and data
formats, along with expertise in the data science tools for upscaling, downscaling,
cross-walking, and harmonizing voluminous and heterogeneous datasets into a single data structure. This process involves ingesting a large number of Level-0 (raw
source) datasets, their transformation into a single large Level-1 (coherent integrated) dataset, and then the application of quality control tools to produce reliable
Level-2 (inspected and quality controlled) datasets. This process is accomplished
using a scientific workflow implemented using Python language, allowing the
reproducibility of the dataset. The resulting dataset size is measured in Petabytes,
and its calculation required high-performance computing (HPC). This data resource
features documentation, a data model, metadata, a codebase, and both publicly
available extracts of the data and also privately controlled source datasets, along
with visualization and data download services.
The FEWSION Database™ 1.0 includes some of the following data types:
• 43 Commodity flow categories, based on the “SCTG+FEWSION” code scheme
– Food and Beverages (for people)
– Agricultural Products
– Fuels (Natural Gas, Diesel, Gasoline, Coal)
– Electricity
– Water Use
– Surface water flows and transfers
– … and all other major commodity types
• Flows between 3,143 US Counties and 8 Foreign Regions
• Seven transportation modes (Pipeline, Power Grid, Rail/Train, Road/Truck,
Water/Ship, Air/Plane, Mixed)
• 2012 annual data
B. Kinne et al.
withdrew at a national level, is one important example. Other actors such as basin
commissions, watershed associations, energy cooperatives, and third- party certification programs can fill the gap with forward-looking initiatives that target specific
problems. For these efforts to bear fruit, policymakers must be willing to listen to the
scientific community, and scientists must be able to communicate the results of their
work in language accessible to policymakers and the general public.
21.4.3 Data Fusion Case Study
The FEWSION™ project (https://fewsion.us) is a data fusion effort funded in 2016
by the National Science Foundation’s interdisciplinary INFEWS program
(Innovations at the Nexus of Food Energy and Water Systems). FEWSION brings
together a large number of academic and government data sources to describe the
commodity flows in the US FEW system in a single seamless dataset. This data
enables place-based researchers and those studying a single component of the system to place their work within the broader perspective of the entire FEW system.
This data fusion requires expertise in a large number of distinct datasets and data
formats, along with expertise in the data science tools for upscaling, downscaling,
cross-walking, and harmonizing voluminous and heterogeneous datasets into a single data structure. This process involves ingesting a large number of Level-0 (raw
source) datasets, their transformation into a single large Level-1 (coherent integrated) dataset, and then the application of quality control tools to produce reliable
Level-2 (inspected and quality controlled) datasets. This process is accomplished
using a scientific workflow implemented using Python language, allowing the
reproducibility of the dataset. The resulting dataset size is measured in Petabytes,
and its calculation required high-performance computing (HPC). This data resource
features documentation, a data model, metadata, a codebase, and both publicly
available extracts of the data and also privately controlled source datasets, along
with visualization and data download services.
The FEWSION Database™ 1.0 includes some of the following data types:
• 43 Commodity flow categories, based on the “SCTG+FEWSION” code scheme
– Food and Beverages (for people)
– Agricultural Products
– Fuels (Natural Gas, Diesel, Gasoline, Coal)
– Electricity
– Water Use
– Surface water flows and transfers
– … and all other major commodity types
• Flows between 3,143 US Counties and 8 Foreign Regions
• Seven transportation modes (Pipeline, Power Grid, Rail/Train, Road/Truck,
Water/Ship, Air/Plane, Mixed)
• 2012 annual data
B. Kinne et al.