Toward verifiable science assessment reporting The Global Change

Toward verifiable science assessment reporting: The Global Change Information System (GCIS) MPI-M Seminar – September 17, 2014 Peter Fox + (RPI) - GCIS Semantics Lead, pfox@cs. rpi. edu, @taswegian, http: //tw. rpi. edu + lots of others (esp. R. Wolfe, C. Tilmes, X. Ma) www. globalchange. gov

Overview • U. S. National Climate Assessment • About the GCIS – Who are we? – What did we do and why? – Underlying methods and technologies – What are our plans for the future? • Sneak peak of more verifiable science… 2

U. S. Global Change Research Program The Program: • Coordinates Federal research to better understand prepare the nation for global change • Prioritizes and supports cutting edge scientific work in global change • Assesses the state of scientific knowledge and the Nation’s readiness to respond to global change • Communicates research findings to inform, educate, and engage the global community

Global Change Research Act (1990), Section 106 …not less frequently than every 4 years, the Council… shall prepare… an assessment which– • integrates, evaluates, and interprets the findings of the Program and discusses the scientific uncertainties associated with such findings; • analyzes the effects of global change on the natural environment, agriculture, energy production and use, land water resources, transportation, human health and welfare, human social systems, and biological diversity; and • analyzes current trends in global change, both human- induced and natural, and projects major trends for the subsequent 25 to 100 years. 4

National Climate Assessments Climate Change Impacts on the United States (2000) Global Climate Change Impacts in the United States (2009) Climate Change Impacts in the United States (2014) See: http: //globalchange. gov/ 5

NCA 2009 http: //nca 2009. globalchange. gov 6

Outline for Third NCA Report • • Letter to the American People Executive Summary: Report Findings Introduction Our Changing Climate Sectors & Sectoral Cross-cuts Regions & Biogeographical Cross-cuts Responses – Decision support – Mitigation – Adaptation • Agenda for Climate Change Science • The NCA Long-term Process • Appendices – Commonly Asked Questions – Expanded Climate Science Info 7

Regions & Biogeographical Cross-Cuts Oceans and Marine Resources Coasts, Development, and Ecosystems

Sectors Water Resources Energy Supply and Use Transportation Agriculture Forestry Ecosystems and Biodiversity • Human Health • • •

Sectoral Cross-Cuts • Water, Energy, and Land Use • Urban Systems, Infrastructure, and Vulnerability • Impacts of Climate Change on Tribal, Indigenous, and Native Lands and Resources • Land Use and Land Cover Change • Rural Communities • Biogeochemical Cycles

globalchange. gov - v 2. 0 11

National Climate Assessment 2014 12

Global Change Information System (GCIS) Long Term Vision: The Global Change Information System (GCIS) is intended to eventually become a unified web based source of authoritative, accessible, usable and timely information about climate and global change for use by scientists, decision makers, and the public. 13

Global Change Information System (GCIS) Long Term Vision: The Global Change Information System (GCIS) is intended to eventually become a unified web based source of authoritative, accessible, usable and timely information about climate and global change for use by scientists, decision makers, and the public. Initial Prototype: Coincident with the release of the Third National Climate Assessment (NCA) or May 6 2014, the GCIS supports the distribution, presentation and documentation needs of the NCA, integrating that content into the USGCRP web site and demonstrating the potential for GCIS to support the long term vision. 14

Information Quality Act • • Reproducibility means that the information is capable of being substantially reproduced, subject to an acceptable degree of imprecision. For information judged to have more (less) important impacts, the degree of imprecision that is tolerated is reduced (increased). With respect to analytic results, "capable of being substantially reproduced'' means that independent analysis of the original or supporting data using identical methods would generate similar analytic results, subject to an acceptable degree of imprecision or error. Transparency is not defined in the OMB Guidelines, but the Supplementary Information to the OMB Guidelines indicates (p. 8456) that "transparency" is at the heart of the reproducibility standard. The Guidelines state that "The purpose of the reproducibility standard is to cultivate a consistent agency commitment to transparency about how analytic results are generated: the specific data used, the various assumptions employed, the specific analytic methods applied, and the statistical procedures employed. If sufficient transparency is achieved on each of these matters, then an analytic result should meet the reproducibility standard. " In other words, transparency and ultimately reproducibility - is a matter of showing how you got the results you got. http: //www. cio. noaa. gov/services_programs/IQ_Guidelines_011812. html 15

Complete Traceability for NCA Content Transparency ------------------------------------ Reproducibility Traceable Data Traceable Processes • References • Image sources • Data sources • Link to datasets • Complete metadata • Description of methods • Access to process info & review Traceable Tools • Access to computer code • Description of systems and platforms Easier. . . Harder Traceable Sources

Data and The National Climate Assessment The Challenge • • More than 250 named authors (>1000 contributing!) 827 pages 43 Chapters and Appendices 284 Figures More than 600 Images 3395 References Approximately 83 data sources used across as many as 235 instances 17

Data and National Climate Assessment The Solution • Defined categories of information within the report: – Figure – Image – Data Source • Build a process for collecting source information that will satisfy IQA and HISA requirements: – Named sources and contacts for every figure, image, and data source – Web-based survey that requests inputs that address transparency and reproducibility and build a foundation for providing the Metadata ISO 19115 standard – IT infrastructure that connects and promotes automation between the web-based survey, a structured data server (SDS)/GCIS, and publication to an official, interactive NCA web site 18

The use case-driven iterative approach 19 More details at: http: //tw. rpi. edu/web/doc/TWC_Semantic. Web. Methodology

Ontology engineering use case The first use case • Title: Find data used to generate a report figure • Actor and system: A reader of the National • 20 Assessment Climate Flow of interactions: A reader wishes to identify the source of the data used to produce a particular figure in the NCA. A reference to the paper in which the image contained in this figure was originally published appears in the figure caption. Clicking that reference displays a page of metadata information about the paper, including links to the datasets used in that paper. Pursuing each of those links presents a page of metadata information about the dataset, including a link back to the agency/data center web page describing the dataset in more detail and making the actual data available for order or download.

Ontology engineering use case The first use case 21 • Title: Find data used to generate a report figure • Actor and system: A reader of the National Climate Assessment • Flow of interactions: A reader wishes to identify the source of the data used to produce a particular figure in the NCA. A reference to the paper in which the image contained in this figure was originally published appears in the figure caption. Clicking that reference displays a page of metadata information about the paper, including links to the datasets used in that paper. Pursuing each of those links presents a page of metadata information about the dataset, including a link back to the agency/data center web page describing the dataset in more detail and making the actual data available for order or download.

An intuitive concept map of the 1 st use case 22

An intuitive concept map of the use case Classes and properties recognized from the use case 23

An intuitive concept map of the use case From an intuitive model to an ontology: (1) A defined class or property should be meaningful and robust enough to meet the requirements of various use cases ontologyrecognized can be extended Classes(2) and. An properties from theby useadding case classes and properties recognized from new use cases through the iterative approach 24

Data and The National Climate Assessment The Solution globalchange. gov website NCA Resources Site Web Form ATRAC/XML File Generator Structured Data Server Metadata Entry 25

Dataset metadata from a figure 26

Dataset metadata from a image in a figure 27

The second use case • Title: Identify roles of people in the generation of a chapter in the draft NCA 3 • Actor and system: a viewer of the GCIS website • Flow of interactions: A viewer sees that Chapter 6 (Agriculture) in the draft NCA 3 was written by a group of authors mentioned in a list. On the title page of that chapter the reader can view the role of each author, e. g. , convening lead author, lead author or contributing author, in the generation of this report chapter. • We decided to use the PROV-O ontology to describe this use case 28

The three Starting Point classes in PROV-O ontology and the properties that relate them Source: http: //www. w 3. org/TR/prov-o/ 29

Mapping the use case into PROV-O Author of Chapter 6 in NCA 3 is. A Writing of is. A Chapter 6 in NCA 3 30

Roles of agents in an activity in PROV-O Source: http: //www. w 3. org/TR/prov-o/ 31

Mapping roles of chapter authors into PROV-O is. A Author of Chapter 6 Writing of Chapter 6 in NCA 3 is. A Convening lead author Lead author Contributing author is. A 32

Roles of people in the activity ‘Writing of Chapter 6’ Here only three of the eight authors of this chapter are shown. Each author had a specific role for this chapter.

Re-using existing ontologies for the GCIS ontology By such mappings we can use reasoners that are suitable for the PROV-O ontology, and thus to retrieve provenance graphs from the established GCIS 34

GCIS Structured Data Server • Capture – Obtain from a variety of sources: manual input by trusted parties – support staff, agency partners, data centers; automated harvesting from publishers, agency data centers, etc. • Identify – Assign persistent, resolvable, controlled identifiers to each element. • Organize – Capture, discover and represent relationships between elements, including across various types of elements; across data centers; and across agency boundaries. • Present – Provide machine accessible interfaces to retrieve structured metadata, and to search/data mine it. • Maintain – Develop tools and processes to ensure quality and integrity of database contents over time. 35

Global Change Content Elements • Reports, Figures, Images, Research Papers, Journals, Measurements, Datasets, Instruments, Agencies, Projects, People, Models, Algorithms, … • Findings – “Climate is changing. ” “Sea Level is Rising. ” • Concepts: “Impacts of Climate Change on Human Health” “Adaptation” 36

Machine Accessible Metadata globalchange. gov website NCA Resources Site Web Form ATRAC/XML File Generator Structured Data Server 37

Linked Open Data http: //5 stardata. info 38

Identifier Resolution doi: 10. 5067/MEASURES/GSSTF/DATA 308 A common, persistent, citable reference to that dataset. We build GCIS specific identifiers from those: http: //data. globalchange. gov/doi/10. 5067/MEASURES/GSSTF/DATA 308 Then we can resolve it (with content negotiation) on our site, and link it with identifiers for our other resources, including asserting equivalence and linking with the data center responsible for stewardship and distribution of the actual data. We can also refer and link to other repositories of information about those resources. 39

Content Negotiation http: //data. globalchange. gov/doi/10. 5067/MEASURES/GSSTF/DATA 308 The server response from the URI depends on what you ask for: • A traditional browser will ask for HTML, and receive and render a human readable description of the resource. • Web services can request formal, structured XML or RDF metadata about the resource. Our goal is to provide a curated collection of authoritative global change information, but always link back to the data center or publisher responsible for the long term stewardship of the resource. 40

GCIS Structured Data Server data. globalchange. gov 41

GCIS Database/API RESTful API at data. globalchange. gov URLs correspond to ontology URIs Primary storage : RDBMS (Postgre. SQL) Representation is serialized (for JSON) or used in templates (for Turtle) • Turtle representation is exported into a triple store (Virtuoso) which provides a SPARQL endpoint. • • 42

GCIS Ontology (version 1. 2) (a) Classes and properties representing a brief structure of the NCA 3

(b) Classes and properties relevant to the findings of the NCA 3 and each chapter in it 44

(c) Classes and properties about sensors, instruments, platforms, and algorithms, etc. through which datasets are generated 45

A few classes are asserted as sub-classes of PROV-O classes Full GCIS Ontology documents are available at: http: //tw. rpi. edu/web/project/gcis-imsap/GCISOntology 46

(part of) GCIS Ontology 47 For more info, see http: //data. globalchange. gov

Final output of the GCIS ontology • Ontology documentation – http: //escience. rpi. edu/ontology/GCISIMSAP/2/GCISOntology_v_1_2. htm • Concept map – http: //cmapspublic 3. ihmc. us/rid=1 MCJMLST 01 G 0 CSWH-2 YH 4/GCIS_Ontology_v 1_2. cmap • Ontology RDF serialized in Turtle format – http: //escience. rpi. edu/ontology/GCISIMSAP/2/GCISOntology_v_1_2. ttl 48

Global Change Keywords (GCMD) Sample finding: Certain types of extreme weather events have become more frequent and intense, including heat waves, floods, and droughts in some regions. The increased intensity of heat waves has been most prevalent in the western parts of the country, while the intensity of flooding events has been more prevalent over the eastern parts. Droughts in the Southwest and heat waves everywhere are projected to become more intense in the future. GCMD v 8. 0 • • • ATMOSPHERIC/OCEAN INDICATORS > EXTREME WEATHER > EXTREME PRECIPITATION > PRECIPITATION RATE EXTREME WEATHER > HEAT/COLD WAVE FREQUENCY/INTENSITY NATURAL HAZARDS > HEAT NATURAL HAZARDS > FLOODS, PRECIPITATION > PRECIPITATION AMOUNT PRECIPITATION >RAIN SURFACE WATER > FLOODS ATMOSPHERIC PHENOMENA > DROUGHT, EXTREME WEATHER > EXTREME DROUGHT, NATURAL HAZARDS > DROUGHTS 49

SPARQL Example • http: //data. globalchange. gov/examples • List 10 figures and datasets from which they were derived select ? figure, ? dataset FROM <http: //data. globalchange. gov> where { ? figure gcis: has. Image ? img prov: was. Derived. From ? dataset } limit 10 50

Two Parallel Paths 1. Third National Climate Assessment (NCA 3) Traceable Sources • References • Image sources • Data sources Traceable Data Traceable Processes • Link to datasets • Description of • Complete metadata methods • Access to process info & review Traceable Tools • Access to computer code • Description of systems and platforms 2. GCIS

Two Parallel Paths 1. NCA 3 release Traceable Sources • References • Image sources • Data sources Traceable Data Traceable Processes • Link to datasets • Description of • Complete metadata methods • Access to process info & review 2. Populate GCIS Traceable Tools • Access to computer code • Description of systems and platforms

Data and GCIS The Future globalchange. gov website Structured Data Server 53

Interagency Information Integration GCIS can use relationships between all relevant information about global change across the agencies: o From observations to datasets to research papers to models to analyses to organizations to people to synthesized reports to human impacts. . . o Determine agency interdependencies -- An EPA analysis uses a NOAA model dependent on observations from a NASA satellite. o Can present unique interagency metrics "How many papers referenced datasets from a specific satellite? " o Direct users back to agency data centers for more detailed information and the actual content and data.

GCIS Data Mining Structured information with relationships allows integrated data mining, searching, metrics. o What projects provided data used to produce figures that were referenced in the 2013 NCA section about coastal sea level rise impacts? o Which data centers hold data referenced by papers related to forests in the midwest? o Which agencies have people working on projects related to societal impacts of extreme weather events? o Show me the latest papers about health impacts of air quality in California. Which datasets were used in the analysis of air quality in California?

Schedule 2013 2014 Now (Sep) 2015 2016 Release (5/6) NCA Report Initial data sets Full data sets Indicators Demo Pilot Health Assessment Ontology Improvements Sustained NCA Earth Observation Assessment (possible support) 56

Staff (some of many contributors) U. S. Global Change Research Program (USGCRP), National Coordination Office (NCO): Robert Wolfe 1, Curt Tilmes 1, Steve Aulenbach 2, Brian Duggan 2, Justin Goldstein 2, Amanda Mc. Queen 2, Julie Morris 2, Glynis Lough 2 National Climate Assessment (NCA) Technical Support Unit (TSU): David Easterling 3, Paula Hennon 4, Angel Li 4, April Sides 6, Mark Phillips 5, Sarah Champion 4, Andrew Buddenberg 4, Devin Thomas 6 Habitat Seven (NCA Web Design and Development): Jamie Herring, Phil Evans, Aires Almeida, Graham Blair Rensselaer Polytechnic Institute (RPI) Tetherless World Constellation (TWC) (Semantic Web Information Modeling): Peter Fox, Xiaogang Ma, Patrick West, Stephan Zednik, Jin Zheng Forum One (globalchange. gov Web Design, Development and Integration): Michael Rader, John Schneider, Keenan Holloway, Sarah Le. Nguyen 1. 2. 3. 4. 5. 6. NASA University Corporation for Atmospheric Research NOAA/NCDC The Cooperative Institute for Climate and Satellites (CICS), North Carolina State University National Environmental Modeling and Analysis Center (NEMAC), UNC Asheville ERT, Inc. 57

See also • Ma, X. , Fox, P. , Tilmes, C. , Jacobs, K. , Waple, A. , 2014. Capturing and presenting provenance of global change information. Nature Climate Change. 4, 409– 413. doi: 10. 1038/nclimate 2141 • Tilmes, C. , Fox, P. , Ma, X. , Mc. Guinness, D. , Privette, A. P. , Smith, A. , Waple, A. , Zednik, S. , Zheng, J. , 2013. Provenance representation for the National Climate Assessment in the Global Change Information System. IEEE Transactions on Geoscience and Remote Sensing, 51 (11), 5160 -5168. • Xiaogang Ma, Jin Guang Zheng, Justin C. Goldstein, Stephan Zednik, Linyun Fu, Brian Duggan, Steven M. Aulenbach, Patrick West, Curt Tilmes, Peter Fox 2014, Ontology engineering in provenance enablement for the National Climate Assessment, Environmental Modelling and Software, 16, 191 -205. doi: 10. 1016/j. envsoft. 2014. 08. 002 – Open access (until October 17, 2014): http: //authors. elsevier. com/a/1 Pc 6 G 4 s. Kh. E 0 y 1 E 58

Sneak peek for what is next http: //tw. rpi. edu/web/pro ject/ECOOP 59

Courtesy: C Tilmes Climate Informatics: Human Experts and the End-to-End System, by Rood and Edwards 60 http: //www. earthzine. org/2014/05/22/climate-informatics-human-experts-and-the-end-to-end-system/

Questions and Comments? For more information visit http: //www. globalchange. gov and http: //data. globlchange. gov

Next … i. Python meets NCA=National Climate Assessment Stace Beaulieu