Detection of Clear Air Turbulence by Airborne
Weather Radar using RR-MWF Method
Renbiao Wu
College of Electronic Information and
Automation
Civil Aviation University of China
Tianjin, China
rbwu@cauc.edu.cn
Zhe Zhang
College of Electronic Information and
Automation
Civil Aviation University of China
Tianjin, China
cauc_2012@163.com
Yuandan Fan
College of Electronic Information and
Automation
Civil Aviation University of China
Tianjin, China
13920736311@163.com
Hai Li
College of Electronic Information and
Automation
Civil Aviation University of China
Tianjin, China
elisha1976@163.com
Xiaoguang Lu
College of Electronic Information and
Automation
Civil Aviation University of China
Tianjin, China
xglu@cauc.edu.cn
Abstract—The precipitation of CAT (Clear Air Turbulence)
is lower than that of the convective turbulence, resulting low
SNR echoes are received by airborne weather radars in the
detection of CAT. This will inevitably lead to poor
performance on spectrum width estimation where pulse pair
processing (PPP) method is used. To address this issue, an echo
spectral moments estimation method based on reduced-rank
multistage wiener filter (RR-MWF) is proposed by introducing
space-time adaptive processing algorithm for airborne weather
radar turbulence detection performance enhancement in low
SNR scenarios. The proposed method inherits the capability of
enhancing echo SNR by accumulating signals coherently, both
in the spatial and the temporal dimensions. The adaptive RR-
MWF weighted vector and cost function are constructed under
the MSE (Mean Square Error) criterion, therefore the spectral
moments can be accurately estimated for CAT, which is
considered as one of the distributed weather targets. Numerical
simulations show that the RR-MWF outmatches the PPP
method when SNR is lower than 10dB, therefore
demonstrating its effectiveness in low SNR scenarios.
Keywords—clear air turbulence, spectral moments
estimation, low SNR, reduced-rank multistage wiener filter
I. INTRODUCTION
The clear air turbulence (CAT), which typically occurs in
cloudless sky above 6000m, is difficult to be detected by
currently widely used airborne weather radars for civil
aviation [1]. The aircraft experiences and even causes injury
to passengers and damage to airplane [2]. In case of
encountering with CAT, according to the reports, a heavy
CAT encounter at a flight from Paris to Kunming of China
Eastern Airlines on June 18, 2017, which leads to at least 26
passengers were injured [3]. Researches have suggested that
the occurrence rate of mid-high intensity CAT in winters in
NAT areas will rise to its 40%~170% by 2050, compared
with that of before the industrialization [4][5]. This particular
issue is drawing increasing attention to the aviation industry.
For airborne weather radar, turbulence detection is
implemented by estimating the 2
nd
order moment and 3
rd
order moment (Doppler frequency and spectrum width) of
the radar echoes reflected by weather targets. The turbulence
detection is processed according to the spectrum width of the
radar echoes in the certified airborne weather radar. It is
generally considered that a weather target with velocity
spectrum width of 5 m/s or higher would be considered as
turbulence in aviation [6]. One of the widely used Doppler
processing methods is pulse pair processing (PPP) algorithm
[7]. Airborne weather radar can process radar echoes with
low complexity in real time by using PPP method. However,
the effectiveness of PPP algorithm is only limited in high
SNR scenarios. Unfortunately, the precipitation of CAT
areas is lower than that of the convectively induced
turbulence areas. Hence low SNR echoes are received by the
airborne weather radars during CAT detection, which will
inevitably lead to poor estimations on spectrum width where
PPP method is used [8].
The ground-based Doppler weather radar is capable of
CAT detection in some specific mode. WSR-88D is one of
these Doppler radars and can operate in two surveillance
modes: the “clear air” and the “precipitation” [9]. Wide pulse
length and the slow antenna rotation are introduced in the
VCP31/VCP32, the “clear air” modes, for increasing the
number of sampling, thereby improving the radar's
sensitivity and the capability of detecting weather targets in
clear air [10]. In order to effectively detect the CAT, phased
array antenna can be introduced in airborne weather radar,
which will enhance the SNR when processing weak echo
signals and improve the detection of weather targets in low
SNR scenarios. Phased array radar receives echoes on
multiple antennas simultaneously [11]. Thus, it incorporates
additional spatial sampling competency and inherits the
capability of echo SNR enhancement by accumulating
signals coherently, both in the spatial and temporal
dimensions. A wind speed estimation method for low-
altitude wind shear based on space-time adaptive processing
is proposed in [12]. The velocity then can be effectively
estimated even in low SNR scenarios. However, due to the
additional spatial sampling, the dimension of the signal
covariance matrix is huge in these space-time optimal
processors. As a result, it brings a high computation
complexity in practice and is limited in real-time
applications.
In this paper, a CAT detection method based on reduced-
rank multistage wiener filter (RR-MWF) [13] is proposed, by
introducing space-time adaptive processing. It is an effective
reduced dimension space-time adaptive processing method,
which can reduce the intrinsic computation complexity.
Since the traditional Multistage Wiener Filter (MWF) is only
suitable for point targets, this need to be tailored to employed
in the spatially distributed turbulence targets detection. The
This work was supported by the National Nature Science Foundation
of China (NSFC) under grant U1633106, 61471365, U1733116 and
Foundation for Sky Young Scholars of Civil Aviation University of China.
978-1-5386-4112-5/18/$31.00 ©2018 IEEE