The Geo-Atom for GEOINT

Adoption of the geo-atom can improve the decision support capabilities provided by GEOINT models

about-arrange-atom-159693

By Micah L Brachman, Ph.D., University of Maryland; Zachary Mostowsky, NT Concepts; and Ian Jonesi, BlackSky Corporation

Delivering timely and accurate geospatial intelligence (GEOINT) requires collecting data from multiple sources ranging from satellite imagery to ground reports. This geospatial data is often collected, processed, and shared using different data models, leading to many persistent challenges within the GEOINT Community such as ensuring systems interoperability, performing data fusion, and delivering a common operating picture (COP) to end users. The geo-atom is a geospatial data model that defines an association between a location in space-time and a property.[1] This simple model incorporates both discrete object (i.e., vector) and continuous field (i.e., raster) data, and can be extended to represent nearly any feature or phenomenon on the Earth’s surface.[2] In this article, an emergency evacuation scenario is used to demonstrate how adoption of the geo-atom data model could help address challenges related to collecting, processing, and disseminating GEOINT.

The emergency scenario developed for this article is based on an evacuation of the Houston, Texas, metropolitan area due to a hurricane. While this particular scenario is hypothetical, there is significant historical precedent given the devastation and loss of life previously wrought upon this area by Hurricane Harvey (2017), Hurricane Ike (2008), and Hurricane Rita (2005). The basis for this scenario is a map of hurricane evacuation routes and evacuation zones developed by the Harris County Office of Homeland Security and Emergency Management.[3] The data from this map is fused with other geospatial datasets to create a network science model that can predict areas of traffic congestion, and the geo-atom is used to develop the model and deliver a COP to emergency management decision-makers.

  • This article is part of USGIF’s 2019 State & Future of GEOINT Report. Download the PDF to view the report in its entirety. 

Background

The geo-atom has previously been applied to many geospatial problems relevant to the GEOINT Community. For example, the geo-atom has been used to improve the exploitation of remotely sensed imagery by building image-objects that resemble real-world objects rather than using raw pixels as the basis for classification.[4] Another interesting application of the geo-atom is using it to create standardized cartographic products.[5] Work has been conducted to integrate the geo-atom data model into a four-dimensional model that simulates particle dispersion,[6] but to date, there are no known research efforts to integrate the geo-atom with a network science model. 

Network science is the study of network representations of physical, biological, and social phenomena leading to predictive models of these phenomena.[7] In USGIF’s 2018 State and Future of GEOINT Report, Collins et al. state that “the connection between GEOINT and modeling has emerged as a capability that decision-makers and response teams can rely upon to increase the correctness, reliability, and timeliness of their decisions.”[8] The goal of the following research is to demonstrate how integrating the geo-atom with a network science model can enhance the ability of decision-makers to make timely decisions related to the planning and/or execution of an emergency evacuation. 

The basic design of the research presented is as follows: First, geospatial data from several sources is used to develop a network science model. Next, this model is applied to a section of the road network in the Houston area, and the model results are used to predict areas of traffic congestion during a hypothetical hurricane evacuation. The geo-atom is then used to add new input variables to the network flow model, and new model results are produced that show how factors such as flooding and large debris might affect traffic flow. Finally, the model results are disseminated as geo-atom data and are displayed on several software platforms to demonstrate interoperability. This network science application shows how adoption of the geo-atom can improve the decision support capabilities provided by GEOINT models.

Data

Figure 1. Harris County Hurricane Evacuation Map.

The most important data used in this research is a map of hurricane evacuation routes and evacuation zones developed by the Harris County Office of Homeland Security and Emergency Management. This map is shown as Figure 1.[9]

Several GIS datasets were used as well: a polygon shapefile of ZIP Codes and a polyline shapefile of evacuation routes. An additional dataset of hypothetical real-time flooding data was developed and stored as geo-atom data. 

 

Methods

The most common way to create a network science model is by using nodes to represent a set of discrete objects and arcs to illustrate the connections among these objects. For the emergency evacuation model for Harris County, the arcs represent the roads shown in Figure 1 and nodes represent either the start or end point of an evacuation route or a road intersection. The location and number of people who require evacuation from the area is determined using the attributes of the ZIP Code shapefile, and the distance of each road segment is calculated using GIS. A network science model is solved using these data, and the outputs show where areas of traffic congestion may occur. This network science model was originally developed by Micah Brachman and Suzana Dragicevic.[10]

Incorporating real-time data into models is a challenge both within the GEOINT domain and in the broader domain of geographic information science. By using the geo-atom, new variables that represent real-time hurricane impacts can be seamlessly incorporated into the network science model. The model can then generate new outputs, which show how traffic congestion can change as real-time conditions are taken into account. The specific method for encoding the hypothetical real-time flooding data as a geo-atom is shown below:

g = {p, A, a(p)}

Where:

g = flooding

p = (WGS84 Latitude, Longitude)

A = water depth (cm)

a(p) = 170

An example of record for a flood event is:

flooding = {(30.012, -95.806), water depth (cm), 170}

The hypothetical real-time flooding data was then input as new variables in the network science model. It was assumed several roads would be impassible due to flooding. The network science model was then solved again with these new variable inputs and produced results showing how evacuation traffic congestion could change if areas of the roadway became flooded.

 

Results

Figure 2. Potential areas of traffic congestion in the Houston area during a hurricane evacuation (Image Credit: Micah Brachman).

The outputs from the network science emergency evacuation model are shown in Figure 2. This map shows areas of traffic congestion forecast by the network science model that utilizes the geo-atom to incorporate real-time flood depth data that could be reported by an observer on the ground. Emergency managers could use such results to decide how to re-direct vehicles in response to a hazard such as flooding or to help identify locations for alternative evacuation routes.

There are many different software platforms used within the GEOINT Community, thus data interoperability is essential for providing a COP. By transforming the outputs of the emergency evacuation model into geo-atoms, the data can be easily disseminated and visualized using nearly any platform.

The Way Forward

One common critique of geospatial modeling and simulation is that it can be difficult to translate the results into real-world decision-making. To avoid this and other pitfalls, it is useful to evaluate the modeling approach presented above using the characteristics of a good GEOINT model identified by Collins, et al.:[11]

  • Output that is linked to decision or analytic objectives: The model outputs presented can be used by emergency managers and other decision-makers tasked with managing traffic flow during an emergency evacuation. Incorporating real-time, geo-atom data into the model allows these decisions to change to account for developing conditions on the ground.
  • Consistent, identifiable, and available data: ZIP Code and road network data is available for the entire United States, and open-source road network datasets such as OpenStreetMap have global coverage. The availability of emergency evacuation maps varies, but most major metropolitan areas that are under a consistent threat from one or more hazards have an evacuation plan in place.
  • Ability to assess and compare the impact of inputs: For this particular emergency evacuation model, the geography of the road network, designated evacuation routes, and the location of the people who will evacuate can be considered fixed inputs. The impact of real-time, geo-atom data inputs can therefore be easily assessed by comparing the outputs of an instance of the model that does not account for real-time data to the outputs of an instance of the model that does.
  • Consistent outputs: The network science emergency evacuation model produces a map showing where traffic congestion may occur. The generalized mathematical formula for this model presented by Brachman and Dragicevic can use many different types of road networks and population data inputs and will still produce a traffic congestion map as the output.
  • Ability to assimilate real-time observations: One of the major advantages of the geo-atom is it can be used to represent nearly any type of geospatial data. As new real-time sensors are developed and new sources of geospatial data are leveraged for analyses, adoption of the geo-atom will help ensure these observations can be rapidly incorporated into models that support decision-makers.
  • Ability to produce results for advanced visualization platforms: The flexible and scalable design of the geo-atom ensures it can be used to incorporate new, real-time observations into models as well as to share the model outputs. The simplicity of the geo-atom allows these outputs to be visualized on many different types of platforms, ranging from desktop GIS and remote sensing software to mobile devices.

The research presented in this article demonstrates how adoption of the geo-atom data model can help address many persistent challenges related to collecting, processing, and disseminating GEOINT. While the integration of real-time, geo-atom data into a network science model is limited to a hypothetical hurricane emergency evacuation scenario, there are many avenues for future research.

One such avenue is further exploring the potential for the geo-atom to become a widely used standard for geospatial data. Organizations such as the Open Geospatial Consortium could play an important role in facilitating this. Another is designing and implementing an emergency evacuation decision support system that can use nationally or globally available road network data and directly incorporate real-time data from a variety of sensors. Network science is only one possible application of the geo-atom; additional applications that can support the GEOINT decision-making process abound, including cutting-edge machine learning and artificial intelligence capabilities.

 

  1. Michael F. Goodchild. “Geographical Data Modeling.” Computers & Geosciences, 1992:18(4):401-408.
  2. Michael F. Goodchild, May Yuan, and Thomas J. Cova. “Towards a General Theory of Geographic Representation in GIS.” International Journal of Geographical Information Science, 2007:21(3):239-60.
  3. Harris County Hurricane Evacuation Map, Harris County Office of Homeland Security and Emergency Management. http://prepare.readyharris.org/Evacuation-Map. Accessed May 12, 2018.
  4. Ivan Lizarazo and Paul Elsner. “From Pixels to Grixels: A Unified Functional Model for Geographic-Object-Based Image Analysis.” The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2008:38(4/C1).
  5. Robert G. Cromley, Shuowei Zhang, and Natalia Vorotyntseva. “A Concentration-Based Approach to Data Classification for Choropleth Mapping.” International Journal of Geographical Information Science, 2015:29(10):1845-63.
  6. Anthony Jjumba and Suzana Dragicevic. “Integrating GIS-Based Geo-Atom Theory and Voxel Automata to Simulate the Dispersal of Airborne Pollutants.” Transactions in GIS, 2015:19(4):582-603.
  7. National Research Council. Network Science Committee on Network Science for Future Army Applications. The National Academies Press. Washington, D.C.; 2005.
  8. Brian Collins, Ofer Heyman, Joaquín Ramírez, Trude King, Brad Schmidt, Paul M. Young, KC Kroll, Ryan Driver and Carl Niedner. “Modeling Outcome-Based Geospatial Intelligence.” State and Future of GEOINT Report, The United States Geospatial Intelligence Foundation; 2018.
  9. Image credit: Harris County Office of Homeland Security and Emergency Management preparedness website http://prepare.readyharris.org/Evacuation-Map.html.
  10. Micah L. Brachman and Suzana Dragicevic. “A Spatially Explicit Network Science Model for Emergency Evacuations in an Urban Context.” Computers, Environment and Urban Systems, 2014:44:15-26.
  11. Brian Collins, Ofer Heyman, Joaquín Ramírez, Trude King, Brad Schmidt, Paul M. Young, KC Kroll, Ryan Driver, and Carl Niedner. “Modeling Outcome-Based Geospatial Intelligence.” State and Future of GEOINT Report. The United States Geospatial Intelligence Foundation; 2018.

,

Advancing GEOINT Through Clarity in the Employment Market

Establishing a clear set of pathways for GEOINT education, training, and certification to further the professionalization process

,

Generating Synthetic Data for Remote Sensing

Harnessing the full potential of deep neural networks for image interpretation and analysis

,

Next-Generation GEOINT Systems—Influences and Needs

The roles of data collection, storage, management, analysis, and consumption as factors that drive designs