Search

We propose a lightweight radar that autonomously measures snow depth over sea ice from an unmanned aerial vehicle (UAV). Development of this snow radar and its integration with an octocopter UAV is presented. Field trials of the UAV-mounted snow radar, conducted in Antarctica during the summer season of 2017/2018, are also described. The radar allows measurements of snow depths on sea ice between 10 and 100 cm. Additional reflections due to internal layers within the snow are evident at a few measurement points. The snow radar is evaluated for various flight parameters: stationary; flying at speeds between 1 and 3 m/s, and at heights from 5 to 15 m. Evaluation of snow-depth results indicates that a depth accuracy of ±3.2 cm is achieved with stationary measurements, and of ±9.1 cm with measurements at the various flight speeds. [ABSTRACT FROM AUTHOR]

In order to measure the variance of wind velocity, which is contributed from turbulence, via radar observations, it is necessary to remove the unwanted contribution from strong horizontal velocity components through the finite beamwidth of the radar. This effect is referred to as beam broadening. Although the amount of beam broadening has thus far been calculated based on the approximating assumption that the pattern of the beam is rotationally symmetric and has very low sidelobes, we need to take a more theoretical approach to radar—one that does not have a simple beam pattern like the Antarctic Program of the Antarctic Syowa Station (PANSY) radar (69S, 39E). In this article, we clarify the theoretical relationship in a very simple form between the turbulence spectrum, which is directly associated with the variance of turbulence, two-way beam patterns, and the observed spectrum, using autocorrelation functions (ACFs). The theory is thoroughly universal and applicable to any type of atmospheric radar, such that we can quantitatively analyze radar observation systems. Furthermore, we propose a “debroadening” algorithm based directly on this theory and from calculations of the general maximum likelihood (ML). We performed numerical simulations that validate our theory and the algorithm. [ABSTRACT FROM AUTHOR]