Simulation Environment Created to evaluate Dynamic User Input into Adaptive Scanning Strategy

In the current version of the MC&C (version 4), the radar scans are optimized based on user needs and the evolving weather and radar capabilities. End-user preferences are represented by static multi-attribute utility functions and conflicts for system resources are mediated by static inter user trade-off coefficients.  However, one of CASA's goals is to create the capability for users to change preferences and priorities dynamically through mixed human/machine control ( two-way interface).  This year, we have developed a systems simulation environment, based on the environment developed and used for the Meteorological Command and Control testing and development, that will enable the evaluation of alternative two-way interface concepts that incorporate real-time end user decision making.  In addition, because the IP5 test bed has a limited number of nodes and is dependent on actual weather, the simulation capability also supports testing larger networks with alternative mixed-initiative (e.g. both user and machine) control concepts. In addition, to the current DCAS strategy, and to traditional "sit and spin" scanning, the simulation environment supports a) mixed-initiative control strategy where a simulated user can identify a high priority region for scanning and b) a decision-making context strategy where the end user completes job-relevant tasks in real-time and the MCC uses the decision-making context in the resource allocation process.  

The first experiment in the system simulation environment, focused on the NWS forecaster warning decisions.  When an NWS forecaster issues a tornado warning, he or she begins to focus his attention on the next threat. Therefore, once the forecaster has issued a warning for a particular region, the priority for scanning that region could be lowered so that other regions receive more scanning resources. Using a nine-node network and three weather cases, we tested the impact of four resource allocation strategies (traditional sit and spin, DCAS, DCAS with mixed initiative control, and DCAS with knowledge of existing warnings) on how many surveillance scans were allocated to the area of interest designated in the high priority region of the mixed initiative control condition. The weather cases were designed to test boundary conditions and were:  a) the "Weather Everywhere" scenario, designed to reduce the potential benefits positive impact of adaptive scanning, b) the "Localized in the Target Region" scenario, designed to have weather exactly in the simulated forecaster's high priority region, thus increasing the potential positive impact of DCAS, and c) the "Localized with Resource Contention" scenario, designed as middle ground (hybrid between the two previous scenarios).  Results indicate the variants of adaptive sensing outperform "sit and spin" with localized weather covering a subset of the radar network.  In weather events that span the entire radar network with a single large cell, ‘sit and spin' out performs adaptive sensing.

Three different weather scenarios used in the simulation environment (from left to right) "Weather Everywhere","Localized in Target Region" and "Localized with Resource Contention"