Creating a Phased Array Radar for Avalanche Imaging

Avalanches pose a significant threat to human life and settlements, so studying them is key to formulating risk zones. Previously, validating models in order to predict avalanche behavior was limited by a lack of high-quality field data. Radar sensors can be used to gather field data, but in their current form, they only provide single-dimension range measurements. The transmitter power also limits them to a range resolution in the order of 50 m, which is too coarse to provide a true representation of the avalanche dynamics.

Radar antenna arrays at the VDLS bunker.

This project studied the underlying dynamics of avalanche flows with the aid of a newly developed frequency modulated continuous wave (FMCW) phased array radar. With this unique radar, high-resolution 2D velocity measurements and a fully animated 2D reconstruction of avalanche events could be produced. The project involved several institutions: University of Sheffield, University of Cambridge, and University College London (UCL). At UCL, radar system development and the associated radar signal processing were the focus.

The radar operates in a reinforced concrete bunker at a well-equipped avalanche test site in Switzerland, Vallée de la Sionne (VDLS). The bunker is positioned at the foot of a slope opposing the avalanche track, and provides protection for the radar equipment. The test site is used to study avalanche processes using an array of sensors such as radar, pressure sensors, and acoustic sensors. Avalanches can be artificially triggered for experimental measurements by setting off explosives at the peak of the mountain after heavy snowfall. The site is also prone to natural avalanches that can be measured with instruments that are automatically triggered by acoustic sensors.

Data Acquisition and Radar Control

Several specialized data acquisition systems appeared to meet the design requirements, but a National Instruments solution offered tight enough integration between existing hardware and software. An NI PXIe-1082 chassis, NI PXIe-8130 controller, and an 8-channel, 16-bit NI PXIe-6366 X Series DAQ device were chosen with specifications that met the data throughput and dynamic range requirements. Combined with a fast, solid-state drive from a third party, the system can measure entire avalanche events (expected to last at least two minutes) without data loss or buffer overflows.

System cooperation with NI LabVIEW was vital for system software design. The radar had to operate throughout the winter without fail, so software reliability was critical. The radar also interfaces with a triggering system at the avalanche test site bunker. LabVIEW was configured to detect a trigger and begin data acquisition, and also to control relays to turn on the radar transmitter. The software based on LabVIEW was extensively tested before deployment.

Radar Design

Aftermath of natural avalanche event at VDLS with regions of movement highlighted.

The radar developed at UCL follows an FMCW design, which carries out ranging by mixing the transmitted linear frequency modulated (LFM) signal with the received return signal. This mixing process produces a frequency difference (beat frequency) to extract target range and velocity information. In this case, the radar operating frequency is 5.3 GHz, chosen to illuminate the underlying dense region of the avalanche.

The FMCW radar provides sub-meter range resolution with a relatively low transmitter power because the transmitted signal has a wide bandwidth (200 MHz) and the radar continuously transmits. The mixing process also compresses all the signal energy into a very small bandwidth, which reduces the strain on the receiver acquisition hardware relative to other pulse compression techniques.

The UCL radar has eight receiver channels to provide horizontal resolution for the first time (hence, the radar produces 2D images). The eight receiver antennas are randomly spread across a wide 5.3 m aperture for a horizontal resolution of c. 10 m at 1 km range.

The radar has a maximum range of c. 3 km to image the entire avalanche track. To satisfy this requirement, the NI PXIe-6366 DAQ device was used. It is capable of simultaneously acquiring data on eight channels at the 2-MS/s sampling rate required. Estimates of cross-coupling between the radar transmitting antenna and the nearest receiving antenna means the radar had a dynamic range requirement of c. 80 dB; therefore, the receiver A/D converter required an effective number of at least 14 bits.

Avalanche Measurements

After the system acquires measurements from an avalanche, it processes the recorded radar data entirely offline. To make the processing more memory-efficient (the datasets are very large), the data is split into segments using NI DIAdem and its Visual Basic Script (VBS) functions. DIAdem exports the segmented data in a format that third-party software can read directly. Each receiver channel initially processes collected data separately. By performing a moving-target indication on the data, nonmoving targets can be filtered out, and targets associated with the moving avalanche can be isolated. The intensity of each pixel in the image is proportional to the return signal strength of the moving target.

Using tightly integrated software and data acquisition hardware, measurements of three natural avalanche events were made during the winter. Future plans involve applying techniques to track the avalanche front to measure the avalanche velocity using Doppler information. The velocity estimates will be validated against data collected by other radar instruments buried within the avalanche track. The ultimate goal is to produce 2D animations of entire avalanche events with a frame rate of 50 frames per second.

This article was written by Matthew Ash, University College London, UK, using products from National Instruments, Austin, TX. For more information, Click Here