Name: Sweta Parmar

Master’s Thesis Proposal Meeting

Date: Tuesday, November 13, 2018

Time: 3pm
Location: J.S. Coon Building, Room 150
 
Advisor:
Rick Thomas, Ph.D. (Georgia Tech)
 
Thesis Committee Members:
Rick Thomas, Ph.D. (Georgia Tech)
Bruce Walker, Ph.D. (Georgia Tech)
Karen Feigh, Ph.D. (Georgia Tech)
 
Title: Effects of Probabilistic Decision Aid on Weather-Related Decision-Making of Pilots

Abstract: 

Despite many technical advancements in general aviation weather displays, improvement in training programs of pilots and various decision support systems, pilots often fail to comprehend weather information during hazardous weather conditions close to their flight plan, often within 20 nautical miles (nmi) distance (Federal Aviation Administration (FAA) recommended safety distance to maintain from hazardous weather).The FAA finds that 20% (8657) of general aviation accidents from 2003-2007 have weather as one of the major contributing factors (FAA, 2010). Operations in inclement weather require pilots to make multiple evaluations and decisions including identification of presence of hazard, estimation of proximity of weather, estimation of impact to flight path and taking appropriate actions. Next Generation Radar (NEXRAD) is a tool commonly used to aid decision making that provides information about geographically referenced precipitation activity. Pilots have to understand the current weather conditions and extrapolate the future based on NEXRAD. But NEXRAD does not provide support to the pilots regarding the possible impact of weather on their current flight path. Also, many of the weather forecast products designed to provide this support, such as National Convective Weather Forecast (NCWF), only provide deterministic weather information; leaving the pilots to infer the uncertainty in the meteorological data. The current study investigates a probability-based decision aid integrated with NEXRAD display to support the pilot in understanding the uncertainty in current weather situation and improve decision making. The aid provides the probability of a particular route coming within 20 nmi radius of hazardous weather in the next 45 minutes. The experiment will comprise of three groups supported by varying levels of decision support. The “low” support group will have no decision aid (NEXRAD only condition), while the “moderate” support group will be provided with low-accuracy decision aid, and the “high” support group will be provided with high-accuracy decision aid. The participants in each experimental group will undergo a training phase and a test phase that will contribute towards understanding how learning translates from training to actual operational environment in test phase. I hypothesize that the participants in “high” support condition are going to perform better and will be more calibrated than participants in “low” or “medium” support condition. The expected results will have implications on emphasizing the need of high-accuracy probabilistic decision aids in future general aviation cockpits.