Case Study - Technology Solution - Digital Terrain Refinement

Decision aids are widely used to summarize the impact of environmental factors (terrain, atmosphere, ocean, space) on military operations for decision makers. They are popular because they provide a succinct summary of a complex collection of information. However they are also potentially wildly inaccurate. In spite of the information revolution, our knowledge of the true state of the environment is often very limited, particularly when responding to a rapidly developing military crisis. Available environmental data may consist of a mix of old maps and charts, legacy digital data products, historical climate data, and recent images and other direct or indirect observations.

While all of this information can be made available in a Geographic Information System (GIS), it does not necessarily make a consistent cohesive database. The resolution and accuracy of the environmental data may vary widely over the area of interest. And the environmental factors may have very complex direct and indirect impacts on military operations. Without an understanding of the quality / uncertainty of the different sources of environmental data, and the way that the data quality / uncertainty propagates into the product, it is possible for decision-makers to misjudge the risks and make costly mistakes.

The challenge is exacerbated when automated mission planning systems and decision support systems prepare and use the results of decision aid products. When a human expert is involved in the production of environmental products, he can exercise expert judgment on the validity and usefulness of the data or the results. In contrast, direct implementation of the algorithms that access the environmental data and generate a product provides no opportunity for the system to recognize potential problems and may even limit the ability of a human expert to appreciate the quality of the automatically generated results. It is well documented that humans tend to believe any output generated by a computer, especially when rendered with high resolution color graphics . When inaccurate results look just as good as accurate results there is no way for the decision maker to appreciate the potential risks.

In response to these challenges, IET is actively engaged in research to develop: 1) an innovative capability to quantify the quality / uncertainty of environmental data in a consistent representation of uncertainty using our Bayesian inferencing technology; 2) a capability to integrate environmental data from diverse sources with different data qualities, 3) a capability to propagate knowledge about the environmental data quality through models; and 4) a capability to visualize the resulting uncertainty in the product.

The result is a computational hypothesis space that provides capabilities for advanced reasoning about features distributed in time and space based on diverse sources of information.