Instrumentation & Measurement Magazine 23-2 - 43

Fig. 5. a) Spreading of the embolism in xylem (adapted from [9] with permission, © American Society of Plant Biologists, 2013); b) Ultrasonic emissions sensor,
lab. dehydration setup for detection of cavitation [28] (© IEEE, 2019).

plant-specific relation between εr and VWC. Also, risk of tissue wounding/scarring can occur during installation but is
negligible during measurements (minimal heating due to low
voltage, high frequency). Finally, due to short, picosecondpropagation times, the method requires fast analog front-ends.
For example, 10 ps resolution is typically needed for the typical 1% VWC measurement error. This increases the power
requirements, limiting the field-autonomy.

Capacitive Techniques
Capacitance techniques determine the dielectric permittivity
of xylem by measuring the charge time of a capacitor, in which
xylem is a dielectric. In contrast to the impulse excitation used
for TDR, capacitive techniques operate with continuous excitation. For measurements on xylem, frequencies in range of
50-100 MHz or higher are used, to decouple the relative dielectric constant from the electric conductivity (i.e., dielectric loss),
and thus minimize susceptibility to changes in salinity and
temperature. Sensor design is very similar to a volumetric soil
water content measurement sensor. Reported measurement
accuracy is within ±0.1% relative error [25], at least 10 times
better compared to TDR methods. Results also correlate well
with diurnal sap-flow variations [25].

Ultrasonic Detection of Xylem
Embolism
An attractive non-invasive indirect method for tracking xylem embolism formation [26] (Fig. 5a) is passive ultrasonic
sensing. Formations and collapses of individual cavitation
gas-bubbles in xylem under tension produce short mechanical bursts, which are transferred as mechanical waves through
the plant's tissue to the wooden surface. There, such ultrasonic
acoustic emissions can be detected with passive ultrasonic
contact sensors (Fig. 5b). The wooden surface is prepared
by debarking, flattening and applying the acoustic coupling
agent. Given that xylem conduit sizes, and hence diameters of
April 2020

cavitation-bubbles, are in the order of magnitude of 10-100s
μm, ultrasonic emissions typically occur in the ultrasonic spectrum between 100 kHz to 1 MHz [27].
Most commonly, physiologists borrow resonant piezoelectric PZT disc-sensors originally designed for industrial
non-destructive testing. Although the piezosensors themselves are passive, they require a permanently-powered,
always-listening front-end circuitry to capture intermittently
appearing emissions. Being high impedance piezoelectrics,
sensors require charge preamplifiers with typical gains of 40-
60 dB and at least 16 bits analog-to-digital conversion at 2-5
MHz [28].
A single emission bears very little information on xylem
embolism. Hence, the method requires feature extraction,
storage, and statistical processing of large numbers of emissions. Due to limitations of signal acquisition and processing
equipment, early works focused on simplistic metrics, such
as cumulative emission counts, emission rate or individual
emission waveform features like amplitude, energy, rise-time,
frequency content, etc. [29]. In the context of using the ultrasonic acoustic emissions for irrigation, there are two key
questions. Firstly, which signal features are most closely related to the process of xylem embolization-to determine
the method's specificity? Secondly, which features enable the
detection of embolic symptoms early enough to react-to maximize the method's sensitivity?
To ensure the specificity, a non-trivial signal processing
task is to be able to separate the emissions by their origin: xylem cavitations from other acoustic emissions sources, such as
bark shrinkage, fractures in xylem tissue, capillary action of
free water, etc. [30]. In this context, current research associates
the emissions clustered around the 100-200 kHz frequency
band to be correlated with cavitation [30], verified with x-ray
computerized tomography. However, to truly understand the
cavitation signals, additional work is needed in the domain of
physical modelling of the gas bubble formation process, and

IEEE Instrumentation & Measurement Magazine 43

Instrumentation & Measurement Magazine 23-2

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 23-2