Covariance Analysis of Sensors for Wind Field Estimation by Small Unmanned Aircraft

Measurements of aloft winds and turbulence is a fundamental necessity to many different applications. It has been long recognized that contemporary methods for atmospheric measurements, such as towers, balloons, and manned aircraft each are limited in their ability to make in situ measurements of atmospheric phenomena. Small Unmanned Aircraft Systems (sUAS) provide a promising platform capable of operating in regions difficult for manned aircraft to access. They’re typically much cheaper to construct, operate and maintain than manned systems or larger UAS. This enables application in dangerous areas where a platform may be lost, or the use of many different platforms for better temporal (flight scheduling) or spacial (formation flying) measurements. The current meteorological UAS ecosystem consists of many different platform designs, sensor suites, state estimation methods, and wind estimation algorithms. Examining this body of work reveals that for sUAS there exists a compromise between the accuracy of the sensors, the size of the vehicle, and the complexity of algorithms used to determine observed inertial winds. Therefore, achieving the desired wind measurement resolution and accuracy with small unmanned aircraft requires a careful combination of sensors and algorithms. This work describes the creation of a tool that can be employed to make an informed decision regarding system design. Models for typical sensors used to measure atmospheric conditions and aircraft state are provided along with a detailed aircraft model. A sequential estimator is used to perform covariance analysis on the errors in inertial wind measurement associated with the Tempest UAS system. The results are compared to data gathered by the Tempest during a field campaign to demonstrate the effectiveness of determining anticipated errors from a particular system. The analysis is extended to two popular fielded sensor configurations and a flight pattern commonly used for aloft wind measurements from small unmanned aircraft. Results of this comparison are presented along with future directions for augmenting the utility of the covariance analysis tools.