Towards an Optical Chemometric Sensor for Anti-Icing Agents on
Asphalt Pavement
Benny Thörnberg
1a
, Alex Klein-Paste
2b
and Wei Zhang
1c
1
Department of Computer and Electrical Engineering, Mid Sweden University, Holmgatan 10, Sundsvall, Sweden
2
Department of Civil and Environmental Engineering, NTNU, Trondheim, Norway
Keywords: Vibrational Spectroscopy, DOAS, Differential Optical Absorption Spectroscopy, Chemometric Sensor, NIR
Spectroscopy, Anti-Icing, De-Icing, Winter Maintenance, Runway De-Icing.
Abstract: To ensure traffic safety during winter, chemical agents are typically used for de-icing and anti-icing. Smarter
and more precise dispersion of chemicals, which considers local variations in concentration, has the potential
to reduce the total amount applied. This paper presents a study of an optical chemometric sensor capable of
measuring the NaCl concentration and the weight of the dispersed solution per square meter. The experiment
was conducted in an indoor environment, where seven solutions of tap water and NaCl were poured onto a
diffuse surface made of burned clay. Short-wave infrared light was illuminated onto the surface, and the light
was diffusely reflected into a spectrometer after passing through the liquid layer twice. Absorption in the
liquid layer alone can be extracted by subtracting the background and further modeled using Beer-Lambert's
law. Both the concentration of NaCl and the amount of liquid can be computed by fitting an overdetermined
equation system. Experimental results show a strong correlation between actual and computed concentrations,
as well as between actual and computed liquid quantities. Suppression of ambient light, spectral variations of
asphalt, harsh environments, dynamic range, and signal-to-noise ratio are among the challenges for outdoor
chemometric sensing of asphalt pavements.
1 INTRODUCTION
During the winter in cold climates, road surfaces may
freeze, leading to the formation of ice on roads (Meng
et al., 2022). To ensure traffic safety, chemical agents
such as sodium chloride, calcium chloride, and
potassium formate are used for anti-icing, de-icing,
and preventing snow compaction (Klein-Paste and
Dalen, 2018).
However, these chemicals are expensive, can be
corrosive, and may have significant environmental
impacts (Fay and Shi, 2012). For example, road
salting can increase the total amount of transported
solutes in streams and deteriorate surface water
chemistry (Płaczkowska et al., 2024). In fact, many
de-icing salts marketed as more environmentally
friendly than the commonly used NaCl may actually
have similar or even higher toxicity to zooplankton
species (Szklarek et al., 2022). The corrosion of
a
https://orcid.org/0000-0001-5521-7491
b
https://orcid.org/0000-0002-1655-1099
c
https://orcid.org/0009-0009-8694-9762
roadside infrastructure and vehicles by de-icing
chemicals is also well documented (Shi, Fay et al.,
2009). Meanwhile, climate change is causing more
frequent temperature fluctuations around zero, which
in turn increases the need for anti-icing operations.
After application, maintenance personnel have no
effective means of evaluating how the amount and
concentration of anti-/de-icing agents change over
time. Precipitation can dilute the chemicals, and loss
mechanisms such as spray-off, run-off, and snow
plowing can remove the agents from the roadway
(Lysbakken and Norem, 2011). If the concentration
of a de-icing agent in the asphalt road ice layer is too
low, it increases the risk of vehicle slippage and
traffic accidents (Kurczynski and Zuska, 2022). On
the other hand, excessive use of chemicals can drive
up costs and lead to environmental pollution.
A chemometric sensor that provides data on the
prevailing chemical amount (in typical units such as
Thörnberg, B., Klein-Paste, A. and Zhang, W.
Towards an Optical Chemometric Sensor for Anti-Icing Agents on Asphalt Pavement.
DOI: 10.5220/0013272600003941
Paper published under CC license (CC BY-NC-ND 4.0)
In Proceedings of the 11th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2025), pages 431-438
ISBN: 978-989-758-745-0; ISSN: 2184-495X
Proceedings Copyright © 2025 by SCITEPRESS – Science and Technology Publications, Lda.
431
grams/m² or lbs/lanemile) and concentration (in w%)
would be a valuable tool for operational decision-
makers. It could indicate whether re-application is
necessary or predict the freezing point of the solution.
Smarter and more accurate dispersion of chemicals,
taking local variations in concentration into account,
has the potential to reduce the total amount applied.
This would decrease the environmental burden,
enhance traffic safety, and lower costs.
DOAS (Differential Optical Absorption
Spectroscopy), typically used for atmospheric
modeling in remote sensing applications (Frankenberg
et al., 2005), along with reflectometers using
modulated laser diodes (Yin et al., 2024) and spectral
analysis in transflectance mode of liquid layers on top
of diffuse substrates (Naito et al., 2024), are all
fundamental technologies and potential candidates for
the development of a chemometric sensor.
This paper presents a study of an optical
chemometric sensor capable of measuring both the
quantity and concentration of NaCl. The study is
conducted in an indoor laboratory environment as an
initial step toward developing a vehicle-mounted
sensor for use on asphalt concrete road surfaces. The
research questions addressed in this study are: Is it
possible to measure both the concentration and
quantity of applied NaCl using vibrational
spectroscopy and DOAS in transflectance mode?
What challenges might arise when this measurement
method is applied outdoors on asphalt? What
challenges are associated with selecting optical
components for the sensor?
To the best of our knowledge, no such sensor
technology is currently available on the market or has
been previously published in any scientific literature.
2 RELATED RESEARCH
Various sensors for surface condition monitoring
have been developed based on different physical
principles, which can be divided into two main
categories in terms of measurement methods: direct
and indirect, or contact and non-contact (Homola et
al., 2006).
Non-contact optical measurement methods have
become a research hotspot in related fields due to
their precise detection, convenient installation, large
detection area, and potential for development as
vehicle-mounted sensors (Casselgren et al., 2016),
(Yin et al., 2024), (Tabrizi et al., 2024).
(Peters and Noble, 2019) collected spectral data
for samples of NaCl and KCl in single-salt solutions,
and correlations were developed for differentiating
between solutions of the two species. These
correlations correctly identified the solution type for
all solutions in the test set and estimated their
concentrations with an average error of 0.9%.
(Medina et al., 2022) experimentally investigated
local film thickness and substance concentration
(glycerol in water) in falling films using a non-
invasive multiwavelength measurement technique,
which combines the fluorescence method and near-
infrared image analysis. The film thickness was
evaluated using a VIS camera and high-power LEDs
at 470 nm, while the local glycerol concentration was
determined using a NIR camera and high-power
LEDs at 1050, 1300, 1450, and 1550 nm.
(Naito et al., 2024) developed a novel direct
measurement system using NIR spectroscopy to
monitor the fermentation process of sake mash during
brewing. The framework was reported by (Jiménez-
Carvelo et al., 2019). By proposing the subtraction of
spectra, called differential reflectance, obtained
through two measurements, namely, diffuse reflection
and transflectance, the impact of significant absorption
bands of water and physical properties such as multiple
scattering within the mash was reduced.
(Charpentier et al., 2024) determined the
effectiveness of de-icing products by assessing ice
melt. A combination of infrared thermography and
Raman spectroscopy was used in a laboratory
environment for the measurements.
3 MATERIALS AND METHODS
This chapter describes the materials used, the
experimental setup, and the computational data
analysis in sufficient detail to enable other researchers
to reproduce the presented results.
3.1 Liquid Samples
Seven bottles of liquid with salinity levels ranging
from zero to 20 percent were prepared in accordance
with Table 1.
Table 1: Seven liquid samples having variation in salinity.
Salinity
[w%]
Volume
water [ml]
Weight
water [g]
Weight
salt [g]
Bottle
#
0 300 299.4 0 1
2.0 300 299.4 6.11 2
5.0 300 299.4 15.75 3
10.0 300 299.4 33.27 4
15.0 300 299.4 52.84 5
20.0 300 299.4 74.85 6
1.0 300 299.4 3.02 7
VEHITS 2025 - 11th International Conference on Vehicle Technology and Intelligent Transport Systems
432
The solutions were mixed using tap water and sodium
chloride (NaCl). The tap water used is pure and
recommended for drinking.
Figure 1: Density versus Salinity measured for seven liquid
samples.
Figure 1 shows the density versus salinity for the
seven solutions in Table 1. A 100 ml sample of each
solution was placed on a scale, and the corresponding
densities were calculated. Least-squares fitted first-
order coefficients enable interpolation of any density
within the range of zero to 20 percent.
3.2 Experiment Setup
Optical components were assembled to analyze the
spectral power distribution of light reflected from a
surface. The surface was illuminated using a 35 W
tungsten halogen light source. Figure 2 illustrates the
geometric arrangement of this setup. The
spectrometer probe consists of a collimating lens that
focuses the collected light into an optical fiber, which
is further connected to a spectrometer. Light reflected
from an elliptical area (20 × 30 mm) was collected by
the collimating lens and directed into the optical fiber.
Figure 2: Illustration of the experiment setup.
All dimensions are in mm.
A corresponding photo of the setup is shown in Figure
3. The calibration reference is a 10 mm-thick solid
board made of plastic (PTFE).
Figure 3: Photo of experiment setup.
A tray made of burned clay without surface glaze
was used as a diffuse background for the measure-
ments. The solutions listed in Table 1 were poured into
the tray, creating a thin film of liquid on top of the
diffuse background. The analyzed thicknesses of this
layer were set to 1, 0.5, and 0.25 mm.
Figure 4: Spectral measurement in transflectance mode.
Figure 4 illustrates the measurement mode called
transflectance. The illuminated light is diffusely
reflected by the background while passing twice
through the liquid being analyzed.
The spectrometer (Ocean Insight, Flame-NIR+)
features 128 spectral channels evenly distributed
across wavelengths ranging from 938 to 1664 nm. The
PSNR is 6000:1, and the spectral resolution is 10 nm.
3.3 Calibration Model
The goal of the transflectance measurement is to
quantify the optical transmittance 𝑀
𝜆
in the liquid
layer on top of the background. The wavelength is 𝜆.
The spectrometer is wavelength calibrated, but
not radiometric calibrated. The spectral power
02468101214161820
Salinity [w%]
0.96
0.98
1
1.02
1.04
1.06
1.08
1.1
1.12
1.14
Density vs Salinity
Measured data
First order fit
Towards an Optical Chemometric Sensor for Anti-Icing Agents on Asphalt Pavement
433
distribution of the used halogen light source L
𝜆
must be considered as unknown. The spectral
distribution of reflectivity in the background 𝑀
𝜆
must also be treated as unknown. For that reason, we
need a calibration model that accounts for any impact
from the light source L
𝜆
, the spectrometer
sensitivity C
𝜆
and the background reflectivity
𝑀
𝜆
.
The light intensity WR
𝜆
after reflection in the
calibration reference, having reflectivity W
𝜆
is
described as,
(1)
Reference measurement WR
𝜆
using the
calibration reference is typically done prior to
transmittance measurements. The measured light
intensity from transflectance in the liquid layer on top
of the background is computed as,
(2)
The relative intensity 𝑅𝑅
𝜆
is the ratio between
the intensity of transflectance and the intensity of
light reflected in the calibration reference,
𝑅𝑅
𝜆
=
𝑀𝑅
𝜆
WR
𝜆
=
𝑀
𝜆
∙𝑀
𝜆
𝑊
𝜆
(3)
Similarly, the relative reflection of light in the
background is,
𝑅𝑅
𝜆
=
𝑀
𝜆
W
𝜆
(4)
Characterization of the background 𝑅𝑅
𝜆
is
typically done prior to transmittance measurements.
The transmittance of the liquid layer 𝑀
𝜆
can be
computed as,
𝑅𝑅
𝜆
𝑅𝑅
𝜆
=𝑀
𝜆
(5)
3.4 Beer Lambert’s Law
Beer Lamberts law is a mathematical model for light
passing through an optical medium of depth 𝑑, having
the wavelength dependent mass absorption
coefficient 𝜎
𝜆
and density 𝜌. Transmittance is then
described as,
𝑀
𝜆
=
𝐼
𝜆, 𝑑
𝐼
𝜆
=𝑒
(6)
Absorbance is simply the natural logarithm of the
transmittance,
𝐴
𝜆
=𝐿𝑜𝑔𝑀
𝜆
=−𝜇
𝜆
𝑑
(7
)
The absorption coefficient 𝜇
𝜆
is a description
of light absorption, normalized with respect to depth,
and the mass absorption coefficient 𝜎
𝜆
is normalized
with respect to both depth 𝑑 and density 𝜌.
𝜎
𝜆
=
𝜇
𝜆
𝜌
(8
)
3.5 DOAS Computation
An application of Beer Lambert’s law is made in
Differential Optical Absorption Spectroscopy
(DOAS). Let us consider a mixed liquid of substances
W and S (water and salt), measured at wavelengths λ
1
to λ
N.
Then the absorbance,
𝐴
𝜆
=𝜎
𝜆
𝐶
+
𝜎
𝜆
𝐶
∀𝑖∈1…𝑁
(9
)
This is the fundamental equation for differential
optical absorption spectroscopy (Frankenberg et al.,
2005) and it basically expresses that the total
absorbance is the sum of the absorbance of the
individual substances. Column densities are defined
as C =
𝜌𝑑 [g/cm
2
]. Equation 9 is repeated N times,
such that an overdetermined equation system is
defined for the absorbance 𝐴
𝜆
at a set of spectral
channels at wavelengths 𝜆
to 𝜆
. Column densities
for water 𝐶
, and for NaCl solved in water 𝐶
are the
unknown coefficients, being fitted for least square
minimization. Typically, the computation of pseudo
inverse can be used for the determination of column
densities.
𝜎
𝜆
𝜎
𝜆
𝜎
𝜆
𝜎
𝜆
⋮⋮
𝜎
𝜆
𝜎
𝜆
∙
𝐶
𝐶
=
𝐴
𝜆
𝐴
𝜆
⋮
𝐴
𝜆
(10
)
With known column densities, the salinity can
simply be calculated as,
S=100
𝐶
𝐶
+𝐶
,[𝑤%]
(11
)
Density 𝜌
of the measured solution at the
concentration given by S is determined by using
interpolation in Figure 1. The modelled liquid layer
depth d is computed as,
𝑑=
100 ∙ 𝐶
𝑆∙𝜌
(12
)
WR
𝜆
=L
𝜆
∙W
𝜆
∙C
𝜆
𝑀𝑅
𝜆
=L
𝜆
∙𝑀
𝜆
∙𝑀
𝜆
∙C
𝜆
VEHITS 2025 - 11th International Conference on Vehicle Technology and Intelligent Transport Systems
434
The quantity Q of chemical solutions on the
analyzed surface is computed as,
𝑄= 𝜌
∙𝑑 , [g/cm
2
] (13)
4 RESULTS AND ANALYSIS
Special attention was given to the use of a 35-watt
tungsten halogen lamp. During the initial
experimental work, a suspicion arose that this lamp is
unstable over time. For this reason, an additional
experiment was set up to measure the mean intensity
of an arbitrarily selected wavelength band, 1500 to
1550 nm. The mean intensity was monitored over
time, starting from the moment the lamp was
switched on. Figure 5 shows the intensity from cold
start for three different occasions. All three trials
show the same trend: at least 30 minutes are required
to achieve a reasonably stable illumination. This lamp
was allowed to warm up for two hours before all the
experiments reported in this paper. It may take a few
minutes to conduct a spectral measurement of a liquid
layer, followed by a spectral reference measurement.
This is why the stability of illumination was
important.
Figure 5: Warming up a tungsten halogen lamp is shown as
the mean value of spectrometer output between 1500 and
1550 nm. Three examples of starting a cold lamp are shown.
A one-millimeter-thick layer of tap water was
analyzed according to equations 1 through 6. The
absorption coefficient was computed using equation
7 and plotted versus wavelength in Figure 6. A
comparison was made with published measurements
of water (Palmer and Williams, 1974). The
correspondence is good, except in the region around
1450 nm, where the water molecule exhibits very
strong absorbance.
Figure 6: Absorption coefficient of tap water measured
using a 1 mm thick liquid layer. Comparison is made with
published data (Palmer and Williams, 1974).
Figure 7: Absorption coefficient for 10 w% solution of
sodium chloride, measured at 1, 0.5 and 0.25 mm layer
thicknesses d. Shaded region shows de-selected
wavelengths excluded from DOAS computation.
A similar measurement of a 10 w% solution at
depths of 1.0, 0.5, and 0.25 mm is shown in Figure 7.
The absorption coefficient is not the same for all
depths, even though it ideally should be. Like Figure
6, the region between 1400 and 1510 nm is
inaccurately represented. According to equation 7,
the absorption coefficient μ(λ) of the same solution
should be independent of depth d, which is true for
the range of 950 to 1400 nm. The data around 1450
nm were therefore disregarded for DOAS
computation, as indicated by the shaded area in
Figure 7.
0 102030405060708090100
Time [minutes]
2.55
2.6
2.65
2.7
2.75
10
4
Start of a halogen lamp (1500 to 1550 nm)
First start
Second start
Third start
Towards an Optical Chemometric Sensor for Anti-Icing Agents on Asphalt Pavement
435
Figure 8: Absorption coefficient for seven different
solutions of tap water and sodium chloride NaCl. Shaded
region shows de-selected wavelengths excluded from
DOAS computation.
Figure 8 shows how the absorption coefficient
changes with different salinities. The liquid layer
depth d was held constant at 1 mm. The absorption
coefficient is significantly modulated by changes in
salinity for wavelengths greater than 1500 nm.
Figure 9: Mass absorption coefficient for sodium chloride
NaCl solved in water.
Figure 9 shows the mass absorption coefficient
for sodium chloride solute (NaCl dissolved in water),
𝜎
𝜆
, as defined by equation 8. This information is
typically acquired prior to a salinity measurement,
defined as 𝜎
𝜆
in the DOAS equation (9).
Similarly, for water, 𝜎
𝜆
is also acquired. For
comparison, Li and Brown (1993) published similar
data for salinity in seawater.
Figure 10: Salinity computed with DOAS from measured
spectrum are plotted versus actual salinity for the seven
liquid samples. Layer thickness d was 1 mm.
Figure 11: Quantity Q of solution computed with DOAS
from measured spectrum are plotted versus actual quantity.
Salinity S was held constant at 10 w%.
The computation of DOAS on a 1-mm thick layer
of solution, for the seven different salinity levels
listed in Table 1, is plotted as measured versus actual
salinity in Figure 10. A least-squares line fit is also
shown, indicating a good correlation between the
actual and measured salinity. The actual depth 𝑑 was
kept constant during this measurement, with the
seven DOAS computations reporting the following
depths: 1.0, 1.0, 1.0, 1.1, 1.1, 1.1, and 1.1 mm,
respectively.
The quantity of solution per unit area is more
relevant to report than the layer depth. Another
measurement was performed using a 10 w% solution
at depths of 1.0, 0.5, 0.25, and 0.0 mm. For these
liquid layers, the actual quantity Q of solution was
computed using equation 13 and the density of 1.05
900 1000 1100 1200 1300 1400 1500 1600 1700
Wavelength [nm]
0
2
4
6
8
10
12
Absorption coefficient for 1mm liquid film at varying salinity
tapWater
1%
2%
5%
10%
15%
20%
900 1000 1100 1200 1300 1400 1500 1600 170
0
Wavelength [nm]
-8
-7
-6
-5
-4
-3
-2
-1
0
1
Mass absorption coefficient of NaCl resolved in water
VEHITS 2025 - 11th International Conference on Vehicle Technology and Intelligent Transport Systems
436
g/ml from Figure 1. Measured quantities are plotted
versus actual quantities in Figure 11, along with a
least-squares fitted line. The correlation between
actual and measured quantities is high. Salinity S was
reported as 10.1, 12.4, and 9.4 w% for depths of 1.0,
0.5, and 0.25 mm, respectively. All data points in
Figures 10 and 11 correspond to single
measurements.
5 DISCUSSIONS
The need for measurement technology to assist
maintenance personnel in evaluating the quantity and
concentration of anti-icing and de-icing agents is the
primary motivation for this research. The MD30 from
Vaisala is an example of a vehicle-mounted sensor
designed to measure water layer thickness and
classify road conditions (Tabrizi et al., 2024).
Similarly, Yin et al. (2024) present a laser sensor with
comparable functionality.
The results presented in this paper demonstrate
that DOAS can be applied in transflectance mode to
measure the concentration and quantity of NaCl
solution on an optically diffuse background.
However, the evidence supporting this correlation is
limited due to an insufficient number of measurement
points. The laboratory work required to acquire a
single measurement point was labor-intensive. Future
experimental setups must therefore be automated to
enable the collection of larger amounts of spectral
data. This work represents an initial step toward the
design and evaluation of a vehicle-mounted sensor.
A 35-watt tungsten halogen lamp was used as
active illumination alongside an NIR spectrometer.
The instability of this lamp makes it an unlikely
candidate for vehicle-mounted applications.
However, the wide bandwidth data obtained from the
transflectance measurements could inform the
selection of a set of laser diodes for a mobile
chemometric sensor.
Yin et al. (2024) designed an optical
spectroscopic reflectometer to classify road
conditions using three laser diodes. A similar
arrangement, employing modulated laser diodes,
should also be feasible for measuring the
concentration and quantity of solutions.
The spectral reflectivity of the background was
characterized in accordance with equation 4, which is
an essential component of the data analysis method.
Background calibration could pose a challenge when
applying this method outdoors on asphalt, as asphalt
is a non-homogeneous material with poorly defined
optical properties.
Additional challenges must be addressed for
outdoor implementation, including the effects of
ambient light, relative humidity, and temperature.
Casselgren et al. (2016) specifically discussed a
modulation technique for suppressing ambient light.
Shaikh and Thörnberg (2022) showed that humidity
significantly impacts hyperspectral imaging for
polymer classification. Furthermore, it is well known
that the absorbance of an NaCl solution is
temperature-dependent (Li and Brown, 1993). All
results presented in this paper were obtained in a dry
and stable office environment, with no ambient light
present within the sensitive wavelength region.
Sodium chloride, used in this study, is only one of
many anti-icing and de-icing agents employed for
winter road maintenance. Urea (CO(NH
2
)
2
) is another
example that would likely require a larger optical
bandwidth for accurate measurement. Its stronger
vibrations occur at 1900 and 2200 nm (Susuki et al.,
2018). Detector sensitivity at wavelengths extending
to 2.5 µm would require the use of extended InGaAs
technology, which is typically more expensive than
standard InGaAs for 1700 nm. Additionally, more
expensive materials, such as calcium fluoride (CaF
2
),
might be required for lenses.
6 CONCLUSIONS
A study on an optical chemometric sensor capable of
measuring the concentration and thickness of a
sodium chloride solution layer on a diffuse
background was presented. Differential absorption
spectroscopy using the transflectance mode of
spectral measurements was applied in a laboratory
environment. The correlation between actual and
measured values was found to be satisfactory.
However, due to the limited dataset, the level of
evidence provided by this study is low.
Future work on a vehicle-mounted sensor
involves the design of a spectroscopic reflectometer
based on a limited set of modulated laser diodes. The
experimental setup should be further developed to
enable the scanning of larger pavement areas,
facilitating the collection of larger datasets.
Several challenges were identified, including
background calibration, the effects of humidity,
temperature, ambient light, and the cost of
components.
Towards an Optical Chemometric Sensor for Anti-Icing Agents on Asphalt Pavement
437
REFERENCES
Casselgren, J., Rosendahl, S., Sjödahl, M., Jonsson, P.
(2016). Road condition analysis using NIR illumination
and compensating for surrounding light. Optics and
Lasers in Engineering, 77, 175-182.
Charpentier, C., Brassard, J.D., Marchetti, M., Momen, G.
(2024). Advancing Insight into Runway De-Icing:
Combining Infrared Thermography and Raman
Spectroscopy to Assess Ice Melt. MDPI Applied
Sciences, 14, 5096.
Fay, L. and X. Shi (2012). "Environmental Impacts of
Chemicals for Snow and Ice Control: State of the
Knowledge." Water, Air, & Soil Pollution 223(5):
2751-2770.
Frankenberg, C., Platt, U., Wagner, T. (2005). Iterative
maximum a posteriori (IMAP)-DOAS for retrieval of
strongly absorbing trace gases: Model studies for CH4
and CO2 retrieval from near infrared spectra of
SCIAMACHY onboard ENVISAT. Atmospheric
Chemistry and Physics, 5, 9-22.
Homola, M. C., Nicklasson, P. J., Sundsbø, P. A. (2006) Ice
sensors for wind turbines. Cold Regions Science and
Technology, 46 (2), 125-131.
Jiménez-Carvelo, A. M., Lozano, V. A., Olivieri, A. C.
(2019). Comparative chemometric analysis of
fluorescence and near infrared spectroscopies for
authenticity confirmation and geographical origin of
Argentinean extra virgin olive oils. Food Control, 96,
22-28.
Klein-Paste, A. and R. Dalen (2018). The fundamentals of
Plowing, Anti-icing, De-icing, and Sanding, Chapter 6
in Sustainable Winter Roads Operations. Ed. X. Shi and
L. Fu, Wiley Blackwell.
Kurczynski, D., Zuska, A. (2022). Analysis of the Impact
of Invisible Road Icing on Selected Parameters of a
Minibus Vehicle. Sensors (Basel), 22 (24).
Li, J., Brown, C.W. (1993). Near-IR Spectroscopic
Measurement of Seawater Salinity. Environmental
Science and Technology. 27(8), 1611-1615.
Li, L., Li, H., Tan, Y., Yang, X., Xu, H. (2021).
Investigation of chloride release characteristic of
Chlorine-Based Anti-Icing asphalt mixture.
Construction and Building Materials, 312.
Lysbakken, K. and H. Norem (2011). "Processes That
Control Development of Quantity of Salt on Road
Surfaces After Salt Application." Transportation
Research Record: Journal of the Transportation
Research Board 2258(1): 139-146.
Medina, I., Scholl, S., Rädle, M. (2022). Film Thickness
and Glycerol Concentration Mapping of Falling Films
Based on Fluorescence and Near-Infrared Technique.
Micromachines 2022, 13 (12).
Meng, Y., Zhang, C., Chen, J., Ling, L., Lai, J., Yan, T.,
Wang, Z. (2022). Deicing characteristics and pavement
performance of eco-friendly de-icing asphalt mixture.
Construction and Building Materials, 360.
Naito, H., Doi, R., Maruyama, H., Wada, A., Morio, Y.,
Murakami, K. (2024). Rapid measurement of sake
mash using differential near-infrared spectra of diffuse
transmission and transflectance. J Biosci Bioeng, 137
(4), 274-280.
Palmer, K.F., Williams, D. (1974). Optical properties of
water in the near infrared. Journal of the optical society
of America, 64 (8).
Płaczkowska, E., Kijowska-Strugała, M., Ketzler, G.,
Bogena, H. R., Leuchner, M. (2024). Solute fluxes in
headwater catchments with contrasting anthropogenic
impact. Geomorphology, 454.
Peters, R. D., Noble, S. D. (2019). Using near infrared
measurements to evaluate NaCl and KCl in water.
Journal of Near Infrared Spectroscopy, 27 (2), 147-155.
Shaikh, M. S., Thörnberg, B. (2022). Impact of water
vapour on polymer classification using in situ short-
wave infrared hyperspectral imaging. Journal of
Spectral Imaging. 11.
Shi, X., L. Fay, Z. Yang, A. Nguyen Tuan and Y. Liu
(2009). Corrosion of Deicers to Metals in
Transportation Infrastructure: Introduction and Recent
Developments. Corrosion Reviews. 27: 23.
Susuki, I., Ogawa, M., Seino, K., Nogawa, M., Naito, H.,
Yamakoshi, K., Tanaka, S. (2018). Reagentless
Estimation of Urea and Creatinine Concentrations
Using Near-Infrared Spectroscopy for Spot Urine Test
of Urea-to-Creatinine Ratio. Advanced Biomedical
Engineering. 7, 72-81.
Szklarek, S., Górecka, A., Salabert, B., Wojtal-
Frankiewicz, A. (2022). Acute toxicity of seven de-
icing salts on four zooplankton species– is there an
“eco-friendly” alternative? Ecohydrology &
Hydrobiology, 22 (4), 589-597.
Tabrizi, S. E., Hippi, M., Sullivan, J., Farghaly, H.,
Gharabaghi, B. (2024). Real-Time Monitoring and
Forecasting Ice Layer Thickness Growth Rate and Grip
Loss on Road Network During Winter Storm Events.
Transportation Research Record,
Tian, A., Kang, H., Ma, X., Wang, Y., Ruan, C. (2019).
Road meteorological condition sensor based on multi-
wavelength light detection. In Third International
Conference on Photonics and Optical Engineering.
Wang, F., Lin, H., Xu, P., Bi, X., Sun, L. (2021). Egg
Freshness Evaluation Using Transmission and
Reflection of NIR Spectroscopy Coupled Multivariate
Analysis. Foods, 10 (9).
Xu, H., Zheng, J., Li, P., Wang, Q. (2017). Road icing
forecasting and detecting system. (2017). AIP
Conference Proceedings, Volume 1839, Issue 1.
Yang, H., Chen, J., Cai, X., Greszik, D., Dreier, T.; Schulz,
C. (2014). In Liquid film thickness measurement by
two-line TDLAS, 8th International Symposium on
Measurement Techniques for Multiphase Flows, April
01, 2014; AIP, pp 232-235.
Yin, C., Jianguo, L., Huaqiao, G., Jiaoshi, Z., Xiuli, W.,
Characterization of road surfaces using three near-
infrared diode lasers. (2024). Instrumentation Science
& Technology, 52 (3), 311-349
VEHITS 2025 - 11th International Conference on Vehicle Technology and Intelligent Transport Systems
438
