Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 71
5
A Roadmap for Spatial Data
Infrastructure Implementation
A
well-designed spatial data infrastructure (SDI) for the U.S. Geological
Survey (USGS) that is based on best practices and focused on the agency's
mission will have a high probability of success provided there is adequate
planning and execution. Planning and execution will need to entail both a broad
vision and steps that address specific needs, resources, and organizational structure
in the USGS. On the basis of the experience of other agencies that have imple-
mented SDI programs, the Survey may encounter some challenges as it prepares to
implement an SDI, with challenges ranging from general organizational concerns
to technical considerations. A roadmap tailored to address issues specific to the
USGS can provide the help to ensure that its implementation is successful. This
chapter provides a roadmap of both general guidance and specific actions needed
for SDI implementation that is based on the committee's findings and its vision
for the USGS SDI.
A ROADMAP
A roadmap can serve as a starting point for planning, developing, and
implementing a comprehensive SDI program at the USGS. The committee
proposes a roadmap for SDI implementation in Figure 5.1 and divides SDI
implementation into three cyclical phases: (1) preparation and planning; (2)
design, development, and testing; and (3) rollout and refinement. It proposes
some general steps in each phase to assist the USGS in implementing an effective
SDI. The general steps are described below.
Although the committee proposes a roadmap with phases and steps for the
Survey to consider, it was beyond its task and expertise to provide details of
71
OCR for page 72
72 ADVAnCIng stRAtEgIC sCIEnCE: A sPAtIAl DAtA InFRAstRUCtURE RoADmAP
Phase 1:
Prepare & Plan
Review Best
Practices
Leadership
Assignments
Strategic
Planning
Phase 2:
Design, Develop, & Test
Process
Identification &
Definition of Development
Standards
Software
Development
Pilot Training
Phase 3: Development
Program
Program
Rollout & Refine
Follow-up
Metrics
Organizational
Rollout
Process
Adjustments
Foundation for All Phases
Leadership ˇ Communication ˇ Cultural Change ˇ Resources
IT management ˇ User Priorities ˇ Partnerships
Figure 5.1 Roadmap for implementing the USGS spatial data infrastructure.
specific tasks for carrying out its vision for the SDI. Such a detailed roadmap
will need to be developed within the agency and tailored to the specific needs
and resource limitations of the USGS, and decisions will have to be made by the
organization's leaders who understand its strengths and limitations.
OCR for page 73
A ROADMAP FOR SPATIAL DATA INFRASTRUCTURE IMPLEMENTATION 73
Phase 1: Preparation and Planning
Programmatic preparations and plans are critical in the first phase of SDI
implementation. A first step is the appointment of key leaders for envisioning,
establishing, and carrying out the vision for an effective SDI. One notable
position is that of the SDI program director, who will need to report to the
USGS Directorate level for such an agency-wide effort. The core team taking
on the challenge of initially scoping and envisioning an SDI will need to consist
of talented and skilled professionals that understand the value and purpose of
an SDI and are capable of carrying out the vision through effective project
planning and management. It may be necessary to attract outside professionals
through financial incentives to complement the SDI talent already at the USGS.
The work of the core team will include reviewing similar SDIs--such as those
of the National Aeronautics and Space Administration (NASA), the National
Geospatial-Intelligence Agency (NGA), and the British Geological Survey
(BGS)--for best practices and lessons learned at a higher level of detail than
those already outlined in this report; determining and defining SDI system
requirements on the basis of the six science directions of the USGS Science
Strategy and user needs of other agencies, local governments, academe, and
the public; determining the organizational structure for the SDI; and identifying
goals, establishing timeframes and milestones, and developing performance
metrics. Once the initial planning is complete, it will be important to announce a
general outline for implementing the SDI program because communication and
outreach will play a decisive role in the success of the program.
Phase 2: Design, Development, and Testing
The second phase would entail designing, developing, and testing the SDI
program. One of the first steps to undertake in designing the SDI is to identify and
define standards that are specific to the USGS mission. Once the standards are
determined, the next steps to consider are process identification and development
and software development. The former identifies common and documented
processes that can enable the SDI to function smoothly across the USGS, and
the latter establishes tools for the acquisition, discovery, management, recording,
archiving, and sharing of data. With standards, processes, and software in place
for an SDI prototype, a training development program would be needed to allow
staff to become acquainted with the SDI prototype. The training program will be
crucial for providing technical training and support and for building organizational
support and buy-in at all levels. After the prototype has been introduced, it will
be beneficial to deploy a pilot program on a small scale in the USGS to test how
well the pilot program works and to identify and rectify glitches and incorporate
these improvements into the prototype and education efforts.
OCR for page 74
74 ADVANCING STRATEGIC SCIENCE: A SPATIAL DATA INFRASTRUCTURE ROADMAP
Phase 3: Rollout and Refinement
The third phase in implementing an SDI will need to include a process
for rolling out the SDI throughout the USGS and a process for fine-tuning the
program. To ensure that people are properly informed and trained, an institution-
wide training program will need to be in place before the SDI is unveiled, and
retraining will need to be offered periodically for users to understand the system
better. The SDI program will need to implement follow-up metrics that feed back
into review and planning (Phase 1) to determine how well it is being executed
to meet its strategic goals. On the basis of findings gathered with those metrics,
there will need to be a process whereby adjustments can be made to serve users
and fulfill USGS priorities.
ORGANIZATIONAL CONSIDERATIONS FOR SUCCESSFUL SDI
IMPLEMENTATION
Leadership
A designated project official would need to be identified early in the process
to oversee SDI implementation. The SDI implementation program will require
the efforts of a full-time staff officer, and the USGS could explore whether such
an official should be placed at the executive level so that information would be
clearly articulated both in and outside the Survey.
An oversight body would also be essential in advising the SDI staff official
on strategic goals for SDI development and implementation. Oversight could
take the form of a board of experts and could include key external stakeholders
inasmuch as it would be important to determine how an SDI would best serve the
needs of its users and to solicit input from external stakeholders.
SDI implementation would need to have high priority with USGS leadership.
The SDI staff official would need support from all levels of leadership--from
senior managers to the USGS director--and would need to be given commensurate
authority to develop and deploy standards. Doing that will require a sense of
ownership on the part of all levels of staff and a fundamental shift in corporate
culture that will be partly driven by bringing employee rewards and incentives
in-line with SDI goals.
The relationship between the considerations above and ongoing efforts at
data integration and SDI development at the USGS will need to be determined
early in the process. The National Map currently forms a central part of the
SDI at the USGS, but the USGS also has many other data infrastructures, such
as the National Biological Information Infrastructure, and these multiple SDI's
would need to be incorporated into the coherent SDI envisioned in this report.
The newly created USGS Directorate for Core System Science appears to be a
OCR for page 75
A ROADMAP FOR SPATIAL DATA INFRASTRUCTURE IMPLEMENTATION 75
positive step towards this goal and the organizational considerations above may
be best carried out under that new structure.
Management
Implementing a complex, organization-wide system will require effective
methods in program management. Communicating the goals and progress of
SDI development is a critical component of leading and implementing an SDI
that allows for transparency and accountability by internal and external parties.
It is also a useful way to leverage favorable publicity--publicly committing to
developing an improved and comprehensive SDI and publicizing progress during
implementation.
It will be important to manage expectations at all levels because implementing
an SDI will be disruptive and the process will take time to complete. Thus, it
would be helpful to set priorities among deliverables to identify low-hanging fruit
in order to build on early successes. To militate against back-sliding, it would be
useful to consolidate steps that help with incremental progress.
A corporate information policy will need to be developed to address all
aspects in and feeding into the SDI. If there is already an existing policy, it will
need to be reviewed comprehensively and revised accordingly to accommodate
SDI priorities. Incentives and protocols could be developed and deployed to
encourage adherence to policy.
SDI implementation is a multi-tiered, multi-year process, therefore the
USGS will need to consider a project-management structure that is flexible in
allowing for personnel changes and reassignments. An adaptive-management
approach would allow the USGS to respond to an evolving technological, data,
and application environment. Furthermore, managers involved in implementing
the SDI will need to clearly define current and future job responsibilities for their
staff so that written job responsibilities reflect and comply with the corporate
information policy. A paradigm shift in management practices will need to occur
that embraces integrated team achievement and includes incentives and resources
to encourage USGS scientists to share data. As mentioned in Chapter 2, an extra
layer of accountability would be to include SDI implementation as a component
of annual staff-performance evaluations.
In measuring the progress of SDI development, it would be helpful to
develop realistic benchmarks based on the performance of analogous national
and international organizations. The USGS could consider an international peer-
review process to gauge the progress of SDI implementation.
Cultural Change
Success will take clear vision, clear direction, and, most important, buy-in
from all USGS staff ranging from the responsible officer to the individual
OCR for page 76
76 ADVANCING STRATEGIC SCIENCE: A SPATIAL DATA INFRASTRUCTURE ROADMAP
contributing scientist. Implementation of an SDI is a major program for
establishing a geospatial base for USGS professional staff and outside users,
and it would need to be viewed as such by the Survey. Individual scientists are
typically committed to their research; thus in implementing a new SDI, a similar
level of commitment and shared responsibility is needed from researchers to
format and share data, adhere to standards, and generate metadata. An example of
cultural change that resulted in effective institutional leadership and staff buy-in
is the Six Sigma quality program developed by Motorola (see Box 5.1). Such staff
participation, adherence to new established processes, and instilling commitment
constitute a necessary paradigm shift in the USGS.
Resources
Adequate resources are needed to design, develop, establish, operate,
maintain, manage, and lead an SDI. These resources include personnel who
have the proper training, knowledge, skills, and motivation including sufficient
understanding of hardware, software, data, and support infrastructure (such as
space and electric power) to keep up with expectations and technological changes.
Financial resources will also need to be sufficient for the SDI program, but it
should not be overfunded; overfunding can lead to waste, whereas underfunding
can lead to frustration and the inability to reach goals in a reasonable time. For
example, in implementing its SDI, BGS budgeted one-third of its expenses to the
SDI program (Ian Jackson, personal communication, 2012). A similar percentage
of the USGS budget may not be practical or required, but the BGS experience
indicates that commitment and a substantial budget are necessary. Finally, SDI
implementation will require substantial time and effort.
Partnerships
Partnerships are vital to a successful SDI and can be used to support the
USGS Science Strategy. Partnering can help the USGS to align itself with federal
and international priorities and with mandates to share scientific data. Benefits of
partnerships include resource-sharing, access to high-value information resources
that are not available otherwise, and the ability to network shared assets and
address problems with information that a single agency usually cannot address
(for example, in the case of climate change). In addition to leveraging expertise
and experiences in different fields, such collaborations can also enable partners'
use of each other's existing data and knowledge assets to be repurposed, creating
greater efficiency through sharing of data resources. There is also the ability
to mobilize quickly in response to critical studies and the ability to connect to
domain users via application programming interfaces. Furthermore, partnering
OCR for page 77
A ROADMAP FOR SPATIAL DATA INFRASTRUCTURE IMPLEMENTATION 77
BOX 5.1
Driving Cultural Change: Implementation of Six Sigma
Quality at Motorola
In the 1970s, Motorola embarked on a process to improve product
quality by implementing a statistical process control called Six Sigma; it rep-
resented the goal of minimizing defects in all aspects of a product's develop-
ment (Montgomery and Woodall, 2008). Achieving a level of quality of about
three to four defects per million opportunities required a substantial culture
change in the corporation.
To invoke culture change, Motorola mounted a campaign with the goal of
training and garnering buy-in from every employee (Luther Siebert, personal
communication, 2011). An essential tenet of institutional culture change is
commitment by top management, and the CEO of Motorola embraced the
Six Sigma program. The support of senior managers enabled company of-
ficers to deploy the initiative, and its implementation was reinforced by tying
the success of the initiative with individual performance, ensuring that each
employee had a personal stake in quality. For example, scorecards weighed
quality-performance achievement, and dashboards were developed for report-
ing up the chain of command. In addition, a bonus program, in which a high
percentage of employees participated, was partially tied to quality.
The Six Sigma process was a multi-tiered campaign. The first step was
training the trainer. Departments designated candidates to become experts in
statistical process control and designated them as the Six Sigma "black belts"
who would also be "change agents". Staffing was determined as a percentage
of organization population to ensure consistent saturation of the organization
with program experts. Certification of black belts and entry level green belts
was determined by length of training, examination completion, and practical
project completion. It was conducted through Motorola University, the organi-
zation's employee-training program that established a thorough curriculum in
quality and quality management.
Senior management was then trained in the Six Sigma process so that
they would understand the basic concepts and the benefits of such a change
to the corporation. In addition to accountability for company oversight, the role
of senior management included removing barriers and making sure that suc-
cess was recognized. Lower-level executives and managers were also trained
in courses that included reference textbooks and other materials. Simultane-
ously, training teams traveled to all corporate offices and provided extensive
staff training courses with emphasis on engineering and manufacturing. Six
Sigma training was required for employees, and each department had metrics
and goals for Six Sigma performance. For example, managers were periodi-
cally evaluated on their performance in achieving Six Sigma goals, and train-
ing resources remained available in case a manager required assistance.
As a result of the Six Sigma process at Motorola, there was a gradual
but dramatic elimination of defects and an increase in product quality, and
Motorola eventually became known for its product quality. In 1988, Motorola
was the recipient of the Malcolm Baldridge National Quality Award (given by
the president of the United States) for its processes and documentation to
determine a high-threshold measure of organizational quality.
OCR for page 78
78 ADVANCING STRATEGIC SCIENCE: A SPATIAL DATA INFRASTRUCTURE ROADMAP
and making data publicly available to partners present an opportunity to garner
goodwill and political support.
Partnerships require time and effort to develop and maintain and each partner
will need to reassess full control and ownership; this may cause participants to
shift from their agencies' missions or visions. It will also be difficult initially to
establish and agree on shared domains and standards.
There are five categories of partnerships to consider: (1) strategic partnerships
(with agencies such as NASA, the National Oceanic and Atmospheric
Administration (NOAA), and NSF); (2) data partnerships (with agencies such
as the Census Bureau);(3) standards partnerships;(4) academic partnerships; and
(5) technology partnerships (with the commercial sector). The USGS will need
to develop and expand strategic partnerships with key federal players--NASA,
NOAA, NSF, NGA, the U.S. Department of Agriculture (USDA), Department
of Energy (DOE), and the Department of Health and Human Services (HHS)--
on science data to address common SDI issues; these are strategic in that they
provide authority, expertise, and critical information resources for addressing
the USGS Science Strategy. The Survey will also need to establish strategic
partnerships with selected state, tribal, and local agencies to implement SDI
functions that support effective exchange of spatial data among the local, state,
and federal levels, perhaps starting with hazards or water data. It will also be
important to engage key domestic and international scientific organizations in
domain standards development (such as the Consortium of Universities for the
Advancement of Hydrologic Science, Inc.) and interoperability standards (for
example, OGC and ISO). Many of these partnerships will also increasingly
include volunteered geographic information from the public and that is provided
to the USGS through partners and from the USGS's own volunteered science
data programs. The USGS will need to develop partnerships with the science
community, the private sector, and other federal agencies to address technical
issues of SDI data quality, value-added data, and online services, perhaps through
centers of excellence and working groups.
Federal
Once SDI implementation has begun in the USGS, the SDI will need to be
accessible to external agencies and expand to encompass data contributed by
external partners. USGS partnerships will be important for creating or augmenting
USGS data assets through a collaborative process. Conversely, USGS scientists
provide research support for an extensive array of external agencies while
maintaining interfaces with state agencies that create statewide data content.
Ongoing USGS research support to external agencies is an example of
successful partnerships already under way. The unique combination of
biological, geologic, hydrologic, and mapping programs of the Survey provides
independent high-quality data, research support, and assessments needed by
federal, state, local, and tribal policy-makers, resource and emergency managers,
OCR for page 79
A ROADMAP FOR SPATIAL DATA INFRASTRUCTURE IMPLEMENTATION 79
engineers and planners, researchers and educators, and the public. The USGS
provides applied research and data to agencies such as the Federal Emergency
Management Agency, the Army Corps of Engineers, NASA, NOAA, the U.S.
Forest Service, the National Park Service (NPS), and the Bureau of Reclamation.
The USGS Contaminant Biology Program collaborates with other agencies and
organizations--including the Fish and Wildlife Service, the Bureau of Land
Management, NPS, the National Park Service, U.S. Environmental Protection
Agency (EPA), the Bureau of Reclamation, the Office of Surface Mining, and
other federal, state, and local agencies--in the conduct of its research to improve
understanding of the environmental effects of current and emerging contaminants.
Given the multidisciplinary activities of the USGS in support of research
and of external agencies, digital data will need to be shared within the USGS and
with other agencies. Digital data on various scales will need to be archived so that
they are discoverable by researchers and so that they exist in standard formats
to keep reformatting to a minimum. Both the USGS and states provide data, and
the USGS fills the roles of coordination and dissemination.
States
Cooperation with states will be essential for ensuring that information is
available and usable. States are users and providers of USGS data, so it is
necessary for data to flow in both directions. Small, densely-populated states,
such as Delaware, seldom use USGS datasets, because most of their data needs
are on a fine spatial scale. In those small states, local and state governments often
create datasets, such as high-resolution elevation databases and air photographs,
whereas large states with low population densities, such as Montana, might use
USGS datasets. However, state cooperation will be essential for updating the
national-level USGS datasets whether a state is small or large.
Other Researcher Communities
The USGS is in the unique position of being a national provider of spatial
data and the home for a large cadre of scientists and therefore has the potential
to be a more effective national asset of the larger research community. Many
research collaborations already exist with USGS scientists; these should continue,
and new collaborations should be encouraged. It is important that all USGS data
be discoverable, so at a minimum there should be data catalogs that describe
the details of the data that currently exist. The government-wide effort to
catalog spatial data on-line through Data.gov1 includes a large number of USGS
spatial datasets and this effort provides a basis for a future catalog that is fully
integrated into the USGS SDI. Online catalogs of USGS data can foster research
collaborations with other research communities and ensure that data are used by
as many researchers as possible to answer research questions that are important
1http://www.data.gov.
OCR for page 80
80 ADVANCING STRATEGIC SCIENCE: A SPATIAL DATA INFRASTRUCTURE ROADMAP
to the nation. Therefore, datasets need to be discoverable, and resources will need
to be allotted specifically for data-sharing.
Public
The general public is a user of USGS products, and public interest in USGS
products has increased as new ways of producing, displaying, and distributing the
results of USGS data have evolved. The USGS will need to consider the public's
needs as it develops methods to ensure that the SDI provides public access to its
vast array of high-quality products. Access can be provided through portals or
by making USGS data accessible to others for distribution through other popular
access capabilities. The ability to ingest volunteered geographic information
from the public will become increasingly important.
Implementation Partnering
To ensure the broadest utility of the SDI, the USGS will need to consider
implementing it by using open standards and making use of existing infrastructure.
Previous SDIs typically made use of Geographic Information System (GIS) tools
to manage and display data, and users of GIS tools typically need training and
need to license or buy GIS software. Open standards for description and display
of geologic and other Earth science data exist. If open standards are embraced,
USGS data can be used and fused with other data for both traditional and new
applications; this would result in a much wider use of the data than if they were
represented in proprietary or internally developed formats and systems.
Adopting open standards for implementing its SDI will enable the USGS
to leverage and partner with other agencies and organizations effectively. For
example, NASA, NSF, EPA, and USDA collect or disseminate geospatial data. In
addition, adoption of open standards will allow informal or implicit partnerships
to be established without the need for formal and often expensive partnerships.
For instance, ground-based field data that are collected by the USGS, USDA,
or NSF can easily be used to validate or calibrate data collected from the air
or from space, such as data on forest height or speciation observed with radar,
light detection and ranging, or spectral imagers. In another example, crustal
deformation data collected by NSF's Plate Boundary Observatory can be fused
with data collected by NASA through uninhabited aerial vehicle synthetic
aperture radar and paleoseismic and seismic data collected by the USGS to
improve earthquake understanding and forecasting.
TECHNICAL CONSIDERATIONS
Information Technology Management
Information technology (IT) is a fundamental aspect of the research
environment, and the concept of an SDI is merging with similar concepts, such
as "eScience" and "eResearch", used in the scientific community. Those all point
OCR for page 81
A ROADMAP FOR SPATIAL DATA INFRASTRUCTURE IMPLEMENTATION 81
to the important role of defining a "cyberinfrastructure" for undertaking research,
collaborating, sharing data and documents, collecting sensor data, and archiving
for provenance and long-term preservation. For a science organization like the
USGS to support its diverse science workflows would require an evaluation
of current IT infrastructure to ensure that it is aligned with the USGS Science
Strategy.
The USGS will need to implement a comprehensive, long-term knowledge-
management infrastructure that supports end-to-end spatial data management,
including the collection, integration, maintenance, and delivery of multidisciplinary
scientific data. To carry that out, the USGS would need to
ˇ Identify data assets most critical for supporting the Science Strategy and
give high priority to making them discoverable and interoperable.
ˇ Implement structural and syntactic interoperability of USGS knowledge
resources, starting with the highest-priority data assets, followed by semantic
interoperability.
ˇ Develop an effective data-archiving strategy to ensure a persistent and
cumulative knowledge resource.
ˇ Initiate a corporate process to create a comprehensive and consistent
inventory and ensure the quality of the data and create standard metadata for it.
ˇ Make "discovered" data structurally (and eventually semantically)
interoperable so that they may be shared and integrated with other datasets in the
USGS, elsewhere in the United States, and internationally.
ˇ Implement internationally agreed-on data standards.
The Web as a Computing Platform
The USGS lacks a robust data-management structure that can manage and
process knowledge on a continental or global scale. Such a system will need
to serve as a foundation for the next generation of knowledge-driven services
and applications. As mentioned in Chapter 3, several other governments and
scientific organizations have recognized the necessity of layering infrastructure
and have begun to develop systematic and general approaches to their SDIs
with architecture that can scale into the future. The USGS will need to be
supported by an adequate cyberinfrastructure that comprises both hardware
(computing resources, data centers, and high-speed networks) and software tools
and middleware. It will also require an SDI that can help to align the germane
scientific research literature with research data. Additional IT developments in
cloud-based computing infrastructures, semantic technologies, database-centric
computing, and data integration on the Web are platform components that warrant
further evaluation as the USGS embarks on its SDI development (See Box 5.2).
It is beyond the scope of this committee to draft a system design for the
USGS SDI, but the agency can embark on the process by learning from others. It
could establish an IT management team that solicits requirements from the new
OCR for page 82
82 ADVANCING STRATEGIC SCIENCE: A SPATIAL DATA INFRASTRUCTURE ROADMAP
Box 5.2
Information Technology Platforms to Consider for the Spatial
Data Infrastructure
Cloud-Based Computing Infrastructures -- Cloud-based computing in-
frastructures enable groups to host, process, and analyze large volumes of
multidisciplinary data. Such consolidation and hosting help to minimize many
organizational and technical barriers that impede multidisciplinary data-sharing
and coordination. Cloud hosting can also facilitate long-term data preservation,
a task that is challenging for universities and government agencies and is critical
for conducting longitudinal experiments. Provisioning key core geospatial datasets
on the cloud would increase the value of the data and facilitate exploitation by
a large scientific community. Scientific organizations that currently manage their
own scientific cyberinfrastructure will probably turn to cloud-based hosting as a
more efficient alternative. Cloud-based computing infrastructures (such as that
of Amazon.com) are being complemented by a new generation of data-intensive
computing services, such as Google's MapReduce, Apache'sTM HadoopTM, and
Microsoft's Dryad. However, a "smart" SDI still does not exist for automatic data
discovery, acquisition, organization, analysis, and interpretation for information
that is available on the Internet or on individual researchers' hard drives.
Semantic Technologies -- Semantic technologies will need to be considered
as an SDI creates necessary links between various disciplines that use different
jargon if information is be interlinked as part of a global network of facts and
processes. The use of semantic technologies is gaining traction in scientific and
engineering communities, including the life sciences, health care, ecology, marine
science, and computer science. With increasing volume, complexity, and hetero-
generation of researchers and from external partners and research organizations.
The IT management team could also solicit expertise and best practices from
existing commercial software, hardware, and service vendors.
Enterprise Data Management
The USGS will need to implement enterprise data-management techniques
in order to
ˇ Create a robust data-management structure that enables individual
scientist to make their data interoperable.
ˇ Generate linkages between spatial and nonspatial data collections (such
as spreadsheets, published reports, documents, photographs, and engineering
drawings).
ˇ Link spatial and nonspatial data to a unified model of a given environmental
system (such as watershed, ecosystem, and regional climate).
ˇ Direct substantial attention to maintaining appropriate privacy, security,
and provenance.
OCR for page 83
A ROADMAP FOR SPATIAL DATA INFRASTRUCTURE IMPLEMENTATION 83
geneity of data resources, scientists are turning to semantic approaches (in the
form of ontologies--machine encoding of terms, concepts, and relations among
them) to interconnect the different sources of data. The interoperable exchange
of scientific data requires scientific communities to explore common vocabularies
for capturing facts and information specific to their domains of expertise. In that
way, the world's data become joined together on the Web as a common intercon-
nected database.
Database-Centric Computing -- In database-centric computing, computa-
tions are brought to the data rather than the data being brought to the computa-
tions. This will become increasingly important as datasets increase in size beyond
terabytes. The volume of data that Earth scientists will work with can be daunting,
and new approaches to scientific discovery, collection, and analysis will increase
the volume of data. A new generation of parallel database systems from Oracle,
Teradata, and others leverage "MapReduce" as an effective data-analysis and
computing paradigm that performs computations as close to the data as possible
(Dean and Ghemawat, 2004).
Data Integration on the Web -- SDI architects will need to determine the best
way to aggregate huge amounts of semantically rich information and consider
how the resulting information is generated and analyzed. It will be important to
consider such requirements now to support higher-level knowledge-generation
processes of the future. Cloud computing services currently focus on offering a
scalable platform for computing, and future services will need to be built around
the management of knowledge and reasoning over it. Services such as OpenCyc
(www.opencyc.org), Freebase (www.freebase.com), and Wolfram/Alpha (www.
wolframalpha.com) demonstrate how facts can be recorded in such a way that
they can be combined and made available as answers to users' questions.
Priorities for Users
Internal Use and Needs
Internal useonly databases that can be discovered and shared effectively
within the USGS will continue to be needed and the proprietary nature of the
information means that such databases will not be discoverable by external
parties until there is official agreement about the broader accessibility of data to
external agencies and the public. Given the multidisciplinary research activities
of the USGS, digital data will need to be made available on various scales,
archived so that they are discoverable by USGS researchers, and exist in standard
formats so that reformatting is kept to a minimum when they are made accessible
to outside users.
Application Services to Engage and Support Scientific Efforts
An SDI will need to create a foundation for a more open approach to
application services that would promote greater transparency, easier integration,
OCR for page 84
84 ADVANCING STRATEGIC SCIENCE: A SPATIAL DATA INFRASTRUCTURE ROADMAP
Box 5.3
Key Application Services
Data Transformation -- Data transformation services convert data between
different formats and coding systems, between different spatial coordinate systems
and projections, or between different levels of aggregation or disaggregation. Such
services include certain types of image processing services that manipulate images
from remote sensing instruments or aerial photography, geoparsing and geocoding
services that convert location-based information (such as place names, addresses,
and administrative and postal codes) to and from spatial coordinates, gridding algo-
rithms that convert vector-based to raster-based data formats, and semantic transla-
tion services that facilitate interpretation of data between languages, disciplines, and
applications.
Data Integration -- Data integration services typically build on data transforma-
tions to support the assembly of data from different sources (for example, instru-
ments, models, and disciplines) into a combined dataset or database or other derived
product, such as a map. They can include linking services to identify data that have
overlapping geographic coordinates or similar geographic features or characteristics,
alignment services that adjust geometric models to improve spatial matching be-
tween different spatial datasets or images, and filtering services that select data for
inclusion or exclusion on the basis of specified constraints and data characteristics.
Spatial and Statistical Analysis Spatial and statistical analysis services facili-
tate assessment of possible relationships within and between geographic distribu-
and greater reuse and repurposing of distributed data and services. Current
approaches to SDI development, such as the Open Geospatial Consortium (OGC)
Web Services Architecture (Whiteside, 2005), support an array of interoperable
services focused on data transformation, integration, analysis and statistics,
modeling, and visualization (see Box 5.3).
The SDI will need to adopt and implement an open Web services architecture
and take a leadership role in establishing a framework for collaborative
applications development and operations. For that to become a reality, the USGS
will need to play a leadership role in establishing key elements of a community
applications platform, including
ˇ Specific open standards needed for data and service interoperability.
ˇ Key reference datasets and parameters that would facilitate the
interoperability and integration of data and services.
ˇ Coordination mechanisms for ensuring development and implementation
of new standards and reference datasets when and where needed.
ˇ Mechanisms for quality control, measurement, and improvement.
OCR for page 85
A ROADMAP FOR SPATIAL DATA INFRASTRUCTURE IMPLEMENTATION 85
tions, often on the basis of integrated data. A major development over the last several
decades has been that of image classification methods that focus on characterizing
and interpreting imagery data, for example, from satellites and aerial platforms.
More recently, a new generation of statistical analysis methods tailored to spatial
data has emerged that can provide new insights about spatio-temporal phenomena
and processes and new methods for transformation, visualization, and prediction.
Modeling Services -- Modeling services usually apply algorithms based on a
combination of theory and statistics to generate estimates of past or current environ-
mental conditions and changes and to project future events, trends, or risks. Models
that have significant spatial dimensions or elements are increasingly prevalent in
virtually all environmental and social science disciplines and many engineering and
public health fields. The USGS has pioneered the development of diverse spatially
enabled models, such as those related to earthquake prediction, hydrological re-
source management, and land-cover change.
Visualization -- Visualization services not only support hypothesis generation,
analysis, and modeling but provide a mechanism for scientists to use in communicat-
ing their findings to other scientists, applied users, and the public. An example of an
online visualization service that is available at the USGS is the Disease Maps Web
site (http://diseasemaps.usgs.gov). It is a simple tool that allows users to see the
spatial distribution of wildlife and zoonotic diseases (such as West Nile virus) in dif-
ferent years at national or state levels. Another USGS Web site, WaterWatch (http://
waterwatch.usgs.gov/), facilitates spatial data visualization and has daily local-level
stream flow data, which can also be accessed through Google Maps. It displays
real-time stream flow and is able to compare with historical stream flow by station.
The Spatial Data Infrastructure as a Workflow Platform
Over the last decade, various scientific programs have begun to incorporate
workflow methods into their best practices. Most of the programs highlighted
in Chapter 3 use workflows to collect raw observation data and make them
available to thousands of researchers worldwide through specialized analytical
and visualization tools (such as National Center for Atmospheric Research and
National Science Foundation collaboratories). Workflow approaches are being
developed by the OGC through its Open Web Services testbeds and through
the Architecture Implementation Pilot of the Global Earth Observing System of
Systems.
The USGS will need to consider implementing high-throughput workflow
processes as a means of advancing the overall scalability of the SDI in support of
the USGS Science Strategy and the SDI's long-term role in geospatial knowledge
capture, preservation, and reuse. Workflow techniques will need to become an
essential technology layer in an SDI--one that enables research on a large
scale by automating complex data preparation and analysis pipelines and by
facilitating cross-disciplinary analysis, visualization, and predictive modeling.
Providing computing, analytics, visualization, and other application processes
as an SDI layer moves them closer to the data and makes it possible to leverage
OCR for page 86
86 ADVANCING STRATEGIC SCIENCE: A SPATIAL DATA INFRASTRUCTURE ROADMAP
unprecedented power and bandwidth to understand and predict the interaction of
natural and human processes better.
CLOSING REMARKS
The intent of this roadmap is not to be prescriptive regarding SDI development
and implementation but to identify keys to success on the basis of lessons learned
from other SDI implementation efforts. The committee recognizes that there
are many feasible ways of implementing an SDI program successfully, but all
will require that an SDI program be properly defined, led, and supported. The
committee believes that such an effort can be best framed by the phrase "discover
and share for the long term", and it hopes that this phrase can become the mantra
for spatial data handling throughout the USGS.
References
Dean, J. and S. Ghemawat, 2004. MapReduce: Simplified Data Processing on Large Clusters, OSDI,
doi: 10.1145/1327452.1327492.
Montgomery, D.C. and W. H. Woodall 2008. An Overview of Six Sigma, International Statistical
Review, 76:329-346.
Whiteside, 2005. OpenGIS Web Services Architecture Description, OGC 05-042r2, 22pgs.
