JASA 
100 
G 
90 
80 
zal 
sol 
59} 
20| 
https:/doi.org/10.1121/10.0043324 
—8— Azimuth Pods (N= 96 ) 
—8—CPP (N= 174 ) 
-e—FPP (N= 259) 
»”» 
_20 
£ 
3 Vi 
2 
o 
N x 
30| 
20} 
10} 
Frequency (Hz) 
‚0 
— 
105 
65 
Wr 
Propulsion Type 
FIG. 9. Fraction of CTV passages that exceed the bulker reference RNL spectrum (200 m, 14 kn) per propulsion type (left) and violin plots of speeds per pas- 
sage per propulsion type (right). Violin plots were created using the VIOLINPLOT-MATLAB toolbox by Bechtold (2016). 
characteristics that are not captured by any single parameter. 
As such, the speed-length scaling proposed by the J-E model 
‘MacGillivray and de Jong, 2021) does not adequately 
explain the observed variability in CTV SLs. However, the 
öbserved variability in RNLs was relatively low, which sug- 
gests that the measured spectrum can serve as viable input 
for numerical models aiming to include transiting CTVs—a 
vessel class that has not yet been considered in previous 
modeling efforts. 
This study followed recommendations of the ISO 17208- 
3 standard for measuring underwater noise levels from ves- 
sels in shallow water (ISO, 2025). SSCI-based propagation 
loss corrections, including frequency-dependent absorption, 
were applied. Despite this, a residual distance dependence at 
high frequencies remains evident in the data (Fig. 7). Varying 
:he sediment type within the SSCI framework did not remove 
:his dependence, indicating that sediment choice alone is 
unlikely to be the dominant cause. We therefore interpret the 
remaining distance dependence as a small residual underesti- 
mation of propagation loss at higher frequencies. Such an 
uınderestimation could arise from additional loss mechanisms 
:hat are not explicitly accounted for in the applied 
propagation-loss models: for example, losses at the sea sur- 
face due to wave roughness or near-surface bubble layers. 
Although the reported SL _measurements are limited to 
13 different vessels, the low variability over the 529 
recorded passages suggests that the mean result can serve as 
a useful estimation for the SL spectrum of transiting CTVs. 
The observed speed distributions were concentrated in the 
'ransiting regime (Table II; see Fig. 9, right panel), and our 
analysis cannot resolve how SLs scale outside transit. Given 
that vessel-specific variability proved to be the main source 
J. Acoust. Soc. Am. 159 (4), April 2026 
of RNL variation, expanding the dataset to include more 
individual CTVs would be necessary to establish a more 
robust and generalizable source model for this vessel type. 
Moreover, RNLs are expected to vary significantly across 
different operational states. Dedicated measurements of 
maneuvering CTVs—particularly while pushing against 
wind turbine foundation for crew transfer—would be neces- 
sary to complement the findings of this study. 
The inclusion of frequencies up to 40KkHz, although 
beyond the formal applicability range of ISO 17208-3, 
proved informative by revealing distinct high-frequency 
emissions in a subset of vessels, which would not have been 
detected in a strictly band-limited analysis and warrant fur- 
ther dedicated investigation. 
When comparing our measurements to the bulker refer- 
ence RNL curve (200m, 14kn), we find that CTVs emit 
comparable or higher RNLs over large parts of the spec- 
trum, particularly between 100 and 500 Hz and above 1 kHz 
(Figs. 4, 5, 8, and 9). 
This aligns with findings from Hermannsen et al. 
(2025), who observed elevated high-frequency noise emis- 
sions from fast vessels, and with Shipton ef al. (2025), who 
showed that small vessels can emit noise levels comparable 
to or even exceeding those of larger commercial ships. 
Because the bulker reference curve is not a regulatory 
threshold, our comparison is not intended as a compliance 
assessment. Instead, it provides a contextual benchmark that 
situates CTV noise emissions relative to those of larger 
commercial vessels. The results show that CTVs can emit 
RNLs comparable to those of much larger ships, particularly 
above 1 kHz, underscoring their relevance for regional noise 
modelling and impact assessment. 
Basan et al. 
3413