Hierarchical Analysis, Reasoning, and Visualization of Events in Space and Time (HARVEST)

The Army's Distributed Common Ground Systems (DCGS-A) promises to unite hundreds of different data sources into a common operational database. What is needed is a means of analyzing many complex, overlapping data sources, particularly those with spatial and temporal elements. Here, we propose Hierarchical Analysis, Reasoning, and Visualization of Events in Space and Time (HARVEST) to carry out spatio-temporal analysis of asymmetric threats. We combine Bayesian hierarchical models with a new class of spatio-temporal covariance functions that allow us to detect hidden spatio-temporal interaction patterns and model asymmetric spatio-temporal processes. Hierarchical modeling gives us the tools needed to combine a large number of distinct data sources, all from different spatial and temporal scales, into a unified model for prediction. But to deal with asymmetric threats we move beyond conventional approaches in order to analyze far more complex spatio-temporal processes than can be detected by traditional statistical methods. And we move cutting-edge theoretical statistics into a modeling framework that can handle complex, messy real-world data. This modeling framework will provide military analysts with tools needed to effectively assess asymmetric threats and to automate the monitoring of incoming data by using dynamically updated models to generate alerts regarding potential threats.