Atmospheric Measurement Techniques (Dec 2020)

Design and field campaign validation of a multi-rotor unmanned aerial vehicle and optical particle counter

  • J. Girdwood,
  • H. Smith,
  • H. Smith,
  • W. Stanley,
  • Z. Ulanowski,
  • Z. Ulanowski,
  • Z. Ulanowski,
  • C. Stopford,
  • C. Chemel,
  • C. Chemel,
  • K.-M. Doulgeris,
  • D. Brus,
  • D. Campbell,
  • R. Mackenzie

DOI
https://doi.org/10.5194/amt-13-6613-2020
Journal volume & issue
Vol. 13
pp. 6613 – 6630

Abstract

Read online

Small unmanned aircraft (SUA) have the potential to be used as platforms for the measurement of atmospheric particulates. The use of an SUA platform for these measurements provides benefits such as high manoeuvrability, reusability, and low cost when compared with traditional techniques. However, the complex aerodynamics of an SUA – particularly for multi-rotor airframes – pose difficulties for accurate and representative sampling of particulates. The use of a miniaturised, lightweight optical particle instrument also presents reliability problems since most optical components in a lightweight system (for example laser diodes, plastic optics, and photodiodes) are less stable than their larger, heavier, and more expensive equivalents (temperature-regulated lasers, glass optics, and photomultiplier tubes). The work presented here relies on computational fluid dynamics with Lagrangian particle tracking (CFD–LPT) simulations to influence the design of a bespoke meteorological sampling system: the UH-AeroSAM. This consists of a custom-built airframe, designed to reduce sampling artefacts due to the propellers, and a purpose-built open-path optical particle counter (OPC) – the Ruggedised Cloud and Aerosol Sounding System (RCASS). OPC size distribution measurements from the UH-AeroSAM are compared with the cloud, aerosol, and precipitation spectrometer (CAPS) for measurements of stratus clouds during the Pallas Cloud Experiment (PaCE) in 2019. Good agreement is demonstrated between the two instruments. The integrated dN∕dlog (Dp) is shown to have a coefficient of determination of 0.8 and a regression slope of 0.9 when plotted 1:1.