Research Agenda

Spatiotemporal Research Agenda: A Living Spatiotemporal I/UCRC Whitepaper

0. Introduction

We live in a four dimensional world with 3D in space and 1D in time. Studying and Understanding the four dimensions integratively could help us better address many local to global challenges and better prepare our future generations to build a more sustainable living environment on our home planet. The NSF Spatiotemporal I/UCRC is envisioned to advance such studies in three aspects: 1) improve human intelligence by developing a set of spatiotemporal thinking methodologies built into K-16 curriculum, 2) advance computer science with new spatiotemporal data structure, algorithms, software and tools, 3) improve human capabilities in responding to grand scientific and engineering challenges. This research agenda evolves with relevant advancements to implement the vision and includes the theoretical, technical, infrastructural, applicational, and educational aspects .

1. Theories

The theoretical aspects will include research on space, time, spatiotemporal integration, their formal description and linkage to other domains and potential information theory adv
ancement

1) Space: Formulate a theoretical framework for 2D, 2.5D and 3D data models, data structures/algorithms, relations and linkage to existing knowledge from (Hagerstrand, Christaller, Von Thunen, Schaefer, Bunge, Tobler, Miller, etc)
2) Time:

a) Represent continuous time using discrete time snapshots and transform between them using mathematical methods
b) Represent events using snapshot (creation, duration, deletion, etc)
c) Time analyses: Markov processes (in RS, land change modeling), Power series, Fourier analysis, a trace of track of data objects, etc.

3) Spatiotemporal:

a) The unique aspects of integrated space and time, and how that would change our matured methods, techs, theories
b) Spatiotemporal thinking/pattern projecting to computing strategies
c) Scale Problem (Clarke and Irmischer 2016): How do we deal with different space-time scales? What are the implications for MAUP (modifiable area unit problem or modifiable volume unit problem) of space-time, and tying these various units to real world use-cases (like sports and weather modeling), could we standardize the scales/resolutions/variable extent?
d) Spatiotemporal objects: What are they? How to model them? What is the ontology of space-time? Are they continuous or not? Do we move from studying object to processes?
e) Spatiotemporal processes: wave propagations, rainfall, heat waves, diffusion, aggregation, erosion, creation, merging, splitting that can be modeled, e.g. with CA?
f) Ontology: should we have its own knowledge or the ones in GIScience & other domains would be sufficient?
g) Formal representation: how to integrate space and time using mathematical equations so the previously mentioned aspects can be formalized and used for formal reasoning.

4) Relationship to other domains/fields (overlap, complements, shared challenges/opportunities)

a) Can ethical, legal, privacy issues in other domains (bio-medical / HIPAA) be used to inform ST Research where loss geo-privacy is a concern.
b) Can mathematics, physics, information science inform our treatment of spatiotemporal data (moving from discrete representations of continuous phenomena to mathematical representations)
c) what unique contributions we can make for other domains, geography, giscience, earth science, space time integration?

5) Are there extensions to existing information theory approaches (Lippitt, Stow and Clarke 2014)

2. Methodologies & Techniques

The technical dimension includes many aspects, such as visualization, analytics, data mining, learning, modeling, simulation, and geocomputation.

● Form and convert a formal spatiotemporal description to computable equations and models
● Spatiotemporal statistics and analyses, e.g.,

○ Time series analysis
○ Spatiotemporal interpolation? linear? Data scale/limitation/error?
○ Validating simulation and prediction with spatiotemporal measurements based on point and satellite
○ Scale and discretization/interpolation

● Simulation and forecasting:

○ Process modeling, What are the space-time dynamics that we can represent with simulation? What data is required to support these simulations?
○ Modeling change & dynamics, uncertainty
○ CA, Markov, ABM.

● Data and model fusion/integration

○ Multiple scale spatiotemporal data fusion, mining in real time
○ Interoperating and integrating models
○ Combine data/decision, high dimensional, unified spatiotemporal datum, data quality, uncertainty/error propagation

● Spatiotemporal AI

○ Would AI be different or easily adopted for spatiotemporal context
○ How to consider spatiotemporal autocorrelation and heterogeneity in AI (machine learning/deep learning models, machine intelligence etc.).
○ Do spatial lag factors and local methods work for ML? (c.f. Kanevski etc all, CNN-LSTM)
○ How to deal with lack of data problem in spatiotemporal AI/ML studies? how could industry giants, e.g., facebook, amazon, help?
○ How do we using spatiotemporal methodologies for inference or prediction?

● Visualization techniques & tools

○ Broadly accessible/distribution/sharing spatiotemporal simulation & visualization (2D/3D/4D/5D or virtual reality): open source/crowdsourcing
○ Visualization for public communication (to policy makers, other scientist, yourself) in space-time?
○ Could spatiotemporal visualization help transform science research into policy making?
○ Story telling using spatiotemporal data for communication

3. Technologies & Infrastructure

The infrastructure support spatiotemporal research and adoption should be based on computing, data, community and knowledge

● Hardware infrastructure (GPU, MIC, FPGA, Cluster, Cloud, Fog, Quantum Computing, etc.)
● Software Software/Tools and solutions
● (Big) spatiotemporal data (sources, collections, standards, fusion, mining, etc.)

○ Network of data sources, IoT
○ Data infrastructure, sharing, standards, fusion, mining
○ Historical long-tail data treatment
○ Broadly accessible spatiotemporal simulation & visualization (2D/3D/4D/5D or virtual reality)
○ Geoprivacy issues, how to anonymize and mask data, re. bio/medical domains and social science
○ Bridging government and industry spatiotemporal data in openness, authoritativeness, integrity and uncertainty
○ PII, plagiarism detection of data, information, and knowledge

● A framework includes infrastructures of storage, data, models, computing, functional tools, standards in an integrative fashion, e.g., Al Gore’s original Digital Earth concept.
● How to integrate existing infrastructural components for implementing such a framework so we can leverage one for all in a shared fashion?

4. Applications

Spatiotemporal applications is key to many science and engineering aspects, such as civil, industrial, environmental, social & behavioral, history and archaeology

● Natural Resource and Disaster: Land use model, other natural phenomena or disasters, public health problems, urbanization, etc.
● Physical phenomena prediction using big spatiotemporal data
● Spatiotemporal computing in other domains (RS, earth science, climate change, land change/ urbanization, tools, etc.)
● Social Sciences

○ Ways to incorporate human’s behaviors/movement
○ Analysis of trajectory bundles from IoT / digital exhaust data
○ Use of cell phone/mobile positioning data for profiling ST Research
○ Human movement – migration, employment, journey to work, recreation, etc. Specialist actions e.g., journey to crime; model human behavior
○ anonymization – privacy, ethics, legal, durable anonymization that leaves data useful (PII), health records, plagiarism,
○ What can we do to correct the biases in social media, and are there ways to connect social media dynamics with well-established social science (e.g. economic models) that tether subset of social media authors to the general population? Do they adhere to supply/demand and reflect the general theories. How can we link social media to the invisible population they represent (i.e. represent the Census)? Can we lay out best practices in this domain? Can we have a few concrete sets of methodologies? We could try to characterize convenience sample? We could send a survey to the people whose tweets we use? Can we establish infrastructure for social media data? An archive of social media for us that is legal and reusable?

5. High Profiled Applications:

Use-cases and demonstrations for practical problems could help drive the understanding, building coherence, and set up roadmap milestones.

● Stories and exemplars of good ST research and applications areas for ST research. Process of identifying scenarios and and data needs to support ST problem solving.
● Planetary Defense

○ Smart search and discovery
○ A PD knowledge base to capture the historical knowledge
○ A visual analytical tool as a decision support system

● illegal poaching of animals (South Africa) Prevention of extinctions

○ video streams
○ geolocation and temporal
○ analyze in real time

● migration of populations

○ urbanization
○ aggregation of historical data
○ integrating real-time data
○ regional or global security problems resulting from this

● Tracking history of geographic objects
● urbanization & security

○ smart cities & secondary cities
○ population and land use change around cities
○ population migration
○ security

■ civil
■ crime, law enforcement
■ terrorist threat
■ health and emerging diseases
■ natural resource security (food, water, ecological)
■ emerging threats from global climate change

○ tools & data, infrastructure to support urbanization & security

■ digital earth
■ smart cities
■ SDI
■ API, IoT data, usable/actionable information,
■ HPC & cloud
■ open data & public participation
■ social media
■ open source hardware, software, sensor network,
■ mobile computing, wearable computing,
■ geomesh, s-t hashing, etc.
■ sentiment analysis, hotspot,
■ data fusion & integration
■ recommendation systems

6. Workforce Training

● Education

○ curriculum & curriculum change and development

■ introduce the concepts of spatiotemporal to high school students, perhaps through summer school/campu and RET
■ How to integrate concepts into new and existing college level curriculum

○ Short courses/summer school
○ K-12 education
○ Where is our body of knowledge for spatio-temporal research, or what subset of that? Might be useful to check out gisbok.ucgis.org. (Note: BoK revision is under way)

● Scholarship (journals, texts, conferences, etc.)
● Job Market (employment) How does it relate to similar “trends” e.g. Data science?
● Formulation and Promotion of a Research agenda

7. Outreach and Collaboration events

● Training
● Symposium
● Workshops
● Special Issue
● Book publication
● Develop a community portal or media to share info

References

1) Clarke, K. C. and I. J. Irmischer (2016). “On scale in space, time and space-time”, in: Scale in Remote Sensing and GIScience Applications,, D.A. Quattrochi, E.A. Wentz, N. Lam, and C. Emerson eds. Boca Raton FL: CRC Press.
2) Lippitt, C. D.; Stow, D. A. and Clarke, K. C. (2014) On the nature of models for time-sensitive remote sensing. International Journal of Remote Sensing. 35, 18, 6815-6841.

Contributors
Chaowei Yang, cyang3@gmu.edu, Geography & GeoInformation Science, George Mason University
David Mills, dam203@txstate.edu, Geography, Texas State University
Keith Clarke,Geography, kclarke@geog.ucsb.edu UC Santa Barbara.
Robert Stewart, stewartrn@ornl.gov, Oak Ridge National Laboratory
April Morton, mortonam@ornl.gov, Oak Ridge National Laboratory
Jesse Piburn, piburnjo@ornl.gov, Oak Ridge National Laboratory
Alexandre Sorokine, SorokinA@ornl.gov, Oak Ridge National Laboratory
Matt Rice, rice@gmu.edu , Geography & GeoInformation Science, George Mason University
Ben Lewis blewis@cga.harvard.edu Harvard Center for Geographic Analysis
James (Jim) Zollweg, jzollweg@brockport.edu, SUNY Earth Sciences
James Pick james_pick@redlands.edu School of Business, University of Redlands
Qian Liu, qliu6@gmu.edu, Geography & GeoInformation Science, George Mason University
Yun Li, yli38@gmu.edu, Geography & GeoInformation Science, George Mason University
Kejin Cui, kcui2@gmu.edu, Geography & GeoInformation Science, George Mason University
Fikriyah Winata, fwinata2@ilinois.edu, Geography and Geographic Information Science, University of Illinois at Urbana-Champaign
Mei Li,mli@pku.edu.cn, IRSGIS, Peking University
Sensen Wu, wusensengis@zju.edu.cn, School of Earth Sciences, Zhejiang University
Xiuping Jia, x.jia@adfa.edu.au, The University of New South Wales, Australia
Huayi Wu, wuhuayi@whu.edu.cn, State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing (LIESMARS), Wuhan University, China
Zhipeng Gui, zhipeng.gui@whu.edu.cn, School of Remote Sensing and Information Engineering, Wuhan University, China
Mehrdad Koohikamali, Mehrdad_koohikamali@redlands.edu, School of Business, University of Redlands
Lara Kamal, lkamal3@gmu,edu, Computer Science and Mathematics,, George Mason University
Yiqing Guo, Yiqing.Guo@student.adfa.edu.au, The University of New South Wales, Australia
Chang Zhao, chang-zhao@uiowa.edu, Department of Geographical and Sustainability Sciences, University of Iowa
Minrui Zheng, mzheng2@uncc.edu, Center for Applied GIScience and Department of Geography and Earth Sciences, University of North Carolina at Charlotte
Wenwu Tang, WenwuTang@uncc.edu, Center for Applied GIScience and Department of Geography and Earth Sciences, University of North Carolina at Charlotte
Long Chiu, lchiu@gmu.edu, Atmospheric, Oceanic and Earth Sciences Department, George Mason Univeristy
Changjoo Kim, changjoo.kim@uc.edu, Geography & GIS, University of Cincinnati
Xuan Shi, xuanshi@uark.edu, Geosciences, University of Arkansas