ArticlePDF Available

Development and application of a real-time water environment cyberinfrastructure for kayaker safety in the Apostle Islands, Lake Superior


Abstract and Figures

Assessing all pertinent environmental variables to categorize a skill level to safely navigate the water environment can be difficult for inexperienced kayakers, especially at a remote site where internet access is limited. A real-time kayaker safety assessment of water environmental conditions at the Mainland Sea Caves of the Apostle Islands National Lakeshore, Lake Superior is achieved. We present a new cyberinfrastructure that provides kayakers with real-time data access and a Safety Index (SI) with consideration of multiple environmental factors to characterize the degree of navigational difficulty for classifying kayaker skill levels. Specifically, radar reflectivity is added to improve forecasts of dangerous conditions caused by convective storms using state-of-the-art weather and wave modeling. Spectral characteristics of surface waves are employed to correlate the occurrences of extreme and freak waves. In addition, unexpectedly dangerous conditions like coastal upwelling and freak wave occurrence due to changing wind directions are considered. A contingency plan is implemented to handle the issue of possibly missing required environmental data. Display of the SI and visualization of other real-time environmental data are communicated by a power-efficient kiosk. Web analytics demonstrates a public interest in real-time water conditions and the need for the on-site kiosk to provide the latest information before kayakers enter the water. The new real-time water environment cyberinfrastructure for kayaker safety in the Apostle Islands, Lake Superior has been successfully operated since 2014.
Content may be subject to copyright.
Development and application of a real-time water environment
cyberinfrastructure for kayaker safety in the Apostle Islands,
Lake Superior
Joshua D. Anderson, Chin H. Wu
Department of Civil and Environmental Engineering, University of Wisconsin-Madison, Madison, WI 53706, USA
abstractarticle info
Article history:
Received 2 April 2018
Accepted 2 July 2018
Available online 13 July 2018
Communicated by Jay Austin
Assessing all pertinent environmental variables to categorize a skill level to safely navigate the water environ-
ment can be difcult for inexperienced kayakers, especially at a remote site where internet access is limited. A
real-time kayaker safety assessment of water environmental conditions at theMainland Sea Caves of the Apostle
Islands National Lakeshore, Lake Superior is achieved. We present a new cyberinfrastructure that provides kay-
akers with real-time data access and a Safety Index (SI) with consideration of multiple environmental factors
to characterize the degree of navigational difcultyfor classifying kayaker skill levels. Specically, radarreectiv-
ity is added to improve forecasts of dangerous conditions caused by convective storms using state-of-the-art
weather andwave modeling. Spectral characteristics of surfacewaves are employed to correlate the occurrences
of extreme and freak waves. In addition, unexpectedly dangerous conditions like coastal upwelling and freak
wave occurrence due to changing wind directions are considered. A contingency plan is implemented to handle
the issue of possibly missing required environmental data. Display of the SI and visualization of other real-time
environmental data are communicated by a power-efcient kiosk. Web analytics demonstrates a public interest
in real-time water conditions and the needfor the on-site kiosk to provide the latest information before kayakers
enter the water. The new real-time water environment cyberinfrastructure for kayaker safety in the Apostle
Islands, Lake Superior has been successfully operated since 2014.
© 2018 International Association for Great Lakes Research. Published by Elsevier B.V. All rights reserved.
Kayaker safety
Apostle Islands
Lake Superior
Safety Index
Kayakingalong the sculpted sea caves amidst a pristine wilderness is
a popular activity at the Apostle Islands National Lakeshore (APIS), Lake
Superior. The Mawikwe (or Mainland) Sea Caves are among the most
visited destinations in the APIS for their natural beauty and accessibility
(Kraft et al., 2007). Water conditions around the APIS can vary dramat-
ically due to rapidly changing winds or fast moving storms (Scott and
Huff, 1996;Anderson et al., 2015). For instance, sudden cold tempera-
tures can appear at the coasts of the APIS when water at depth is
upwelled when warm surface water that is moved offshore by interac-
tions among wind, current, and the Coriolis force (Csanady, 1984;Chen
et al., 2004;Rao and Schwab, 2007). Coastal processes like refraction,
diffraction, or reection around the islands can transform nearshore
waves to form extreme waves (Anderson et al., 2015) or freak waves
(Dean, 1990;Wu and Nepf, 2002;Liu et al., 2010) that are hazardous
to recreational kayakers and have resulted in tragic drowning incidents
near the Sea Caves [Duluth News Tribune, 8/24/2004; 6/26/2007; 09/
12/2010; 06/09/2011]. Accidents often occur when kayakers have
limited knowledge of the water environmental conditions or underesti-
mate the skill level required to navigate safely (Bailey, 2010;Aadland
et al., 2016). In view of the concerns for kayaker safety, there is a need
to provide and assesswater environmental conditions to reveal any hid-
den or unexpected dangers to kayakers of the APIS.
Several indices related to water environmental conditions have been
used to assess safety of small crafts or kayaks. For example, different
levels of indices like wind speed and/or wave height have been applied
in Small Craft Advisory, Small Craft Advisory for Hazardous Seas, Small
Craft Advisory for Rough Bar, etc., for marine warnings (Tew et al.,
2008). For extreme kayaking, the Sea Conditions Rating System
(SCRS), developed by oceanic kayaking experts Eric Soares and Michael
Powers, is a self-assessment scaling measure for scouting sea conditions
and relating them to skill levels (Soares and Powers, 1999). Variables
considered in the SCRS include site geography, wind speed, wave
height, water temperature, weather condition, time of day, etc. While
the assessed levels obtained from the SCRS are comprehensive, there
are a few caveats. First, a kayaker may underestimate the appropriate
class of skill level dueto personal assessment bias in rating multiple en-
vironmental conditions. A kayaker is therefore strongly encouraged to
compare an individual assessment with others. Second, a kayaker may
not be able to obtain water environmental conditions at the destination
Journal of Great Lakes Research 44 (2018) 9901001
Corresponding author.
E-mail address: (C.H. Wu).
0380-1330/© 2018 International Association for Great Lakes Research. Published byElsevier B.V. All rights reserved.
Contents lists available at ScienceDirect
Journal of Great Lakes Research
journal homepage:
due to unavailable data or spatially varying data caused by complex
physical processes. For instance, a fast moving storm that is capable of
generating meteotsunami like waves (Rabinovich, 2009;Bechle et al.,
2016) may reach the kayaker's target location within an hour without
any warning. Sheltering due to islands can cause spatial variability in
wind and wave properties (Stopa et al., 2013;Anderson et al., 2015).
Likewise, coastal upwelling can cool down water temperatures at cer-
tain localized areas while other areas in Lake Superior remain unaf-
fected (Ragotzkie, 1974;Bogrien and Brooks, 1992). Third, a kayaker
may not have access to timely information in the eld due to the lack
of tools to process data for implicit measures or analytics (e.g. wave
steepness or spectral bandwidth of a wave eld) that can be related to
the likelihood of dangerous extreme waves or unexpected freak waves
(Goda, 1970;Naess, 1985;Wu and Yao, 2004;Hultquist et al., 2006;
Anderson et al., 2015). While theSCRS provides an excellent framework
to relate the water environment to kayaking skill levels, logistical dif-
culties in accessing timely information and assessing reliable scaling
for actual water conditions remains a challenging issue for kayakers.
In the last two to three decades, cyberinfrastructure technology has
greatly advanced to serve as an automated tool for providing water en-
vironment conditions. For example, the Great Lakes Coastal Forecasting
System (GLCFS), originally developed by the Great Lakes Environmental
Research Laboratory (Liu et al., 1984;Schwab et al., 1984), was the rst
cyberinfrastructure to provide near real-time forecasting of currents,
temperatures, winds, waves, and ice conditions using offshore buoys
to collect and transmit observations in real-time at multiple locations
around the Great Lakes. These predictions provide timely information
to lake carriers, mariners, port and beach managers, emergency re-
sponse teams, and recreational boaters, surfers, and anglers. However,
the nearest buoy in Lake Superior to the APIS is approximately 65 km
away, which could be too far to provide a reliable assessment of the
nearshore conditions. In addition, the spatial resolutions of the models
used in the GLCFS have yet been ne enough to resolve nearshore
water conditions around the islands in the APIS. As a result, a new
wave model, WAVEWATCH III, was embedded into the Coastal and
Great Lakes Forecast, developed by National Centers for Environmental
Prediction (Alves et al., 2014) and the National Weather Service (NWS)
to provide nearshore zones around the Great Lakes. Currently, the new
Great Lakes Marine Modeling & Analysis Branch (MMAB) operational
wave models provide excellent wave forecasts for mariners in the
APIS ( Nevertheless,
no available nearshore wave observations have been used to validate
the nearshore wave forecast. Anderson et al. (2015) conducted near-
shore wave measurements around the APIS to calibrate a Simulating
WAves Nearshore (SWAN) model and characterized wave climatology
based upon 35 years of data. To date, the nearshore water environmen-
tal factors have yet to be used in rating skill levels for kayakers in the
Great Lakes. Furthermore, there has been no cyberinfrastructure that
can provide timely assessment of the water environment in relation to
skill levels for kayaker navigation safety at the mainland Sea Caves of
the APIS.
The objective of this paper is to develop a cyberinfrastructure that
provides real-time safety assessment of sea conditions for kayakers at
the mainland Sea Caves of the APIS. A Safety Index (SI) with algorithms
to assess the water environment is presented. We document the design
of hardware and software components of a cyberinfrastructure includ-
ing the Real-Time Wave Observation System (RTWOS) for data acquisi-
tion, an Integrated Nowcast and Forecast Operation System (INFOS) for
data storage, management, integration, and computing, and the Real-
Time Wave Kiosk System (RTWKS) for an energy-efcient display at
the remote location. Results of the SI are examined to identify
concerning safety conditions caused by moving storms, extreme
waves, freak waves, upwelling, and changing wave directions. In addi-
tion, we show a display image of the RTWKS that gives an example of
real-time water environmental information including a SI rating, past
measured water and weather conditions, a latest webcam image, and
a radar reectivity animation related to moving storms. Finally we dis-
cuss the validity and adaptability of ratings for SI, a contingency plan
to handle the issue of possibly missing required environmental data
used in the computation of the SI, and website usage statistics of the
cyberinfrastructure developed in this paper.
Site description
The Mainland Sea Caves study site, comprising a 3.5 km stretch of
sheer sandstone cliffs that rise approximately 15 m above the water sur-
face (Kraft et al., 2007), is located at the western edge of the APIS on the
southwestern shore of Lake Superior (Fig. 1). Water depths are typically
25 m at the base of the cliffs. Atop the cliffs is a hiking trail in a decid-
uous forest that is managed by the National Park Service (NPS). Typi-
cally visitors kayak to the Sea Caves from Meyers Beach (Fig. 1), which
is located approximately 2 km southwest from the start of the Sea
Caves with typical round trip duration about 3 h. Meyers Beach is the
main motor vehicle access point for sea kayakers at any time of the
day or year (Fig. 1). During the busy kayak season from JulyAugust
and on weekends until the end of September, staff from the NPS are sta-
tioned at Meyers Beach.
Wind and wave climatology at the study site is dominated by west-
erly winds and wave directions (Anderson et al., 2015). Monthly aver-
aged winds and waves are at their annual low during July and August
(Kraft et al., 2007), but energetic wind and wave conditions can occur
from synoptic scale weather systems over a daily time scale or convec-
tive storms over an hourly time scale. As a result, summer wave heights
can reach as large as 3 m near the Sea Caves (Anderson et al., 2015), and
wind gusts N20 m/s have been measured by the NOAA real-time mete-
orological station DISW3 (Fig. 1), located at Devils Island. At the west-
erly location of the APIS, the study site is sheltered from strong
northeast wind events that dominate the wave climatology at most
other areas in the APIS (Anderson et al., 2015). Meyers Beach is further
sheltered by the shoreline geography when waves approach from the
northeast, which can deceive kayakers that are examining conditions
at Meyers Beach and planning to visit the Sea Caves.
Safety Index
A Safety Index (SI) is devised to be a points-based skill level by con-
sidering multiple environmental factors that characterize navigational
safety for kayakers. A SI of 100 corresponds to the safest conditions. As
the SI decreases, water conditions become more hazardous forkayakers.
Calculating SI involves assigning conditional pointtotals to multiple var-
iables, summing the total (T), and applying the equation SI = 100 T.
Table 1 shows the variables considered in the computation of the SI and
an example calculation. The point weightings of each category were
originally based on the relative danger posed to kayakers. The method
to combine each category into a composite skill level index was deter-
mined by experts and was tested under multiple scenarios in the
ocean (Soares and Powers, 1999). The SI scale is based upon a rescaling
of the original SCRS so that thresholds N70, 5070, 3050, and b30 of SI
correspond to the expert determined skill levels beginner (I), interme-
diate (II), advanced (III) and expert (IV), respectively (Lull, 2013). In ad-
dition, we further modify the SRCS method for breaking waves and add
a new variable of radar reectivity to characterize atmospheric moving
storms. Details of each variable and the associated conditions for point
allocation are described below.
Environmental factors are classied into three categories (Table 1).
Base category is considered to have no direct effect on navigation.
First, swim distance to safety has a maximum distance of 2 km (Fig. 1)
at the study site as the sheer cliffs provide no escape from the water.
Swim distance is assumed constant at 2 km in the calculation of SI
which limits the maximum SI to 90. Second, entering the sea caves is
991J.D. Anderson, C.H. Wu / Journal of Great Lakes Research 44 (2018) 9901001
not presumed in the calculation of SI, but it is communicated to users
that entering the caves reduces SI by 20. Third, water temperature
(WT) does not pose a danger to navigation but is heavily weighted
due to the dangers of hypothermia. Points assigned for WT are equal
to 72 minus the latest measured WT in Fahrenheit. The base category
considered and point weighting are consistent with the original SCRS.
The atmospheric category consists of meteorological variables that
can impact navigation. Point allocation for wind speed is equal to the
surface wind speed in mile/hour. Measured winds at station DISW3
(Fig. 1) are adjusted to a 2 m height using the power-law wind prole
with an exponent of 0.11 (Hsu et al., 1994). For radar reectivity,
point allocations are determined from the base reectivity scans by sta-
tion DLH located in Duluth, MN. Four groups have been designated as 0
(clear), 5 (light), 10 (medium), and 20 (heavy) for ranges b17.5,
17.530, 3050, and N50 dBZ, respectively. The total area covered by
each group is determined, and the maximum group exceeding an area
of 50 km
within a 200 kmradius from the study site is selected. The ra-
dius of 200 km was based on a typical 3 h kayak trip duration and a po-
tential storm translation speed of 65 km/h (Johns and Hirt, 1987;Geist
et al., 2014). Lastly, fog is not measured near enough to the study site to
Fig. 1. Site locations of the RTWOS and RTWKS within the ApostleIslands National Lakeshore, Lake Superior, WI.
Table 1
Categories and point allocation for computation of kayaker Safety Index (SI).
Category Variable Unit Condition Point allocation Example
Value Points
Base Swim distance to safety m 1 pt./200 m to safety 10 2000 m 10
Entering sea caves No, yes 0, 20 No 0
Water temperature (T) °F 72 T040 60 °F 12
Atmosphere Wind speed (W) mph 1 pt. per W 040 15 mph 15
Radar reectivity Clear, light, medium, heavy 0, 5, 10, 20 Clear 0
Fog (visibility) mi No, yes 0, 20 No 0
Wave SWH ft 4SWH 020 3 12
Steepness kSWH N0.25 010 0.3 10
Spectral bandwidth, Q
N2.6 & SWH N1ft 010 2.2 0
Wave direction deg. (|θ
| mod 180)/9 010 45 5
SI = 100-
992 J.D. Anderson, C.H. Wu / Journal of Great Lakes Research 44 (2018) 9901001
allow automation, but the presence of fog is easily assessed by a kayaker
through visible observation. A warning that dense fog decreases the SI
by 20 is thereby suggested. In comparison to the original SCRS, wind
and fog point allocation is consistent, and radar is a new variable.
Wave category is comprised of four surface water wave variables di-
rectly related to ease of navigation. First, a wave height variable is rep-
resented using the spectral quantity of signicant wave height (SWH)
dened as four times the square root of the zeroth order moment of
the wave energy spectrum (Sorensen, 2006). Point contribution to SI
is four times the SWH, which is consistent with the original SCRS. Sec-
ond, wave steepness (or wave shape) variable has been shown to be
closely related to wave breaking (Nepf et al., 1998;Wu and Nepf,
2002), which can affect boat and kayaking navigation. In the original
SCRS, rating for the breaking wave is mainly by eye without a varying
degree of intensities. We use spectral steepness (SS) as an indication
of the steepness of a wave eld. SS is computed as k× SWH, where k
is 2π/Land Lis the wavelength of the spectral peak frequency that can
be obtained using the linear dispersion relation (Sorensen, 2006).
Based upon the recorded data, simultaneous webcam images and mea-
sured SS values have been examined to determine that breaking and
whitecapping commonly occur when k×SWHN0.25. Point totals for
SS are ramped linearly between k× SWH 0.250.3 from 0 to 10 points
to prevent a staircase result in the SI. Third, spectral bandwidth variable
is associated with an unexpected freak wave (Wu and Yao, 2004;Yao
and Wu, 2005), i.e. a dangerous wave form that a kayaker can encoun-
ter. Freak waves are unusually large and steep compared to the back-
ground wave eld (Kharif and Pelinovsky, 2003;Dysthe et al., 2008;
Liu et al., 2010). When a reected wave meets and combines with an in-
cident wave, a dangerous freak wave may result and is known in the
kayakingcommunity as clapotis(Fig. 2).To date, there is no consistent
predictor for freak waves, but spectral bandwidth has been shown to in-
uence occurrence of freak waves (Naess, 1985). The spectral peaked-
ness parameter, Q
,byGoda (1970) has been used as a proxy for
spectral bandwidth. Points towards the SI are ramped linearly from 0
to 10 between the 0.5 (Q
= 2.26) quantile to the 0.9 (Q
quantile of Q
based on historical measurements collected at the sea
caves and only if SWH is above 1 ft. Both SS and Q
were not considered
in the original SCRS method by Soares and Powers (1999),butthepo-
tential 20 point allocation considered here is consistent with the SCRS
point allocation for breaking waves. Lastly, a wave direction variable
can be a factor to increase the likelihood of vessel capsize (Dahle and
Myrhaug, 1995;Wu and Nepf, 2002). Kayakers are advised to orient
their bow towards oncoming wave crests (Lull, 2013), but this is not al-
ways possible when wave propagation is perpendicular to the path of
travel. Points are thereby allocated as linearly increasing from 0 to 10
as wave direction, θ
, changes from parallel to perpendicular to the di-
rection of navigation, θ
, which is taken to be equal to the orientation of
the shoreline. Note that wave direction is not considered in the original
SRCS, but the 10 point allocation is consistent with the category desig-
nated as miscellaneous.
Here, we briey describe the development of a cyberinfrastructure
that includes a Real-Time Wave Observation System (RTWOS) for data
acquisition, an Integrated Nowcast and Forecast Operation System
(INFOS) for data storage, management, integration, and computing,
and a Real-Time Wave Kiosk System (RTWKS) for data visualization
and display in the following.
To obtain nearshore measurements of the wave eld and water tem-
perature, RTWOS is designed and implemented. On the water side,
waves are measured by an Acculevel pressure sensor by Keller
America with built in air tube for atmospheric pressure compensation.
Pressure head is sampled at 10 Hz with an accuracy of ±1.0 cm. Tem-
perature is measured with a T107 thermistor by Campbell Scientic
once every minute with an accuracy of ±0.4 °C. The sensors are at-
tached to an aluminum bar of a cylindrical frame 1.83 m above the bot-
tom to improve measurements of wave induced pressure (Bishop and
Donelan, 1987). Fig. 3 shows the frame placed ~35 m from the cliff
face at a water depth of ~5 m. Camouaged cabling set along the cliff
face connects the instrument frame to a base station set back from
view on the cliff top. On the land side, a weather-resistant enclosure
Fig. 2. Webcam image taken from RTWOS captured a clapotis freak wave. Kayakers with kayaks approximately 34 m in length have been superimposed on the left for scale purpose.
993J.D. Anderson, C.H. Wu / Journal of Great Lakes Research 44 (2018) 9901001
at the clifftop base station houses electronics for real-timesampling and
transmission of data back to the INFOS at the College of Engineering at
University of Wisconsin, Madison, WI. Specically, a CR800 datalogger
by Campbell Scientic is scheduled to sample data, and transmission is
accomplished via a GX400 cellular modem by Sierra Wireless. Images
of water surface characteristics are captured using a Mobotix M12
ethernet camera that is aimed towards the west. Because the power
requirement of the M12 camera is 3.6 W and approximately equal to
all other electronics combined, we use a programmable switch on the
CR800 to optimize power consumption. Power for the system is
sustained with a 200 Ah battery capacity and 50 watt solar panel. Over-
all, the water and land side components of RTWOS have been operating
since 2011 in collaboration with the NPS at Apostle Islands National
Lakeshore. Archive data including nearshore wave and water tempera-
ture measurements are available for modeling calibrationand data ana-
lytics. Deployment and retrieval of the cylindrical frame with the
sensors of RTWOS takes place in early June and late September.
There are two main components of the INFOS. First, a Central Con-
trolling Computer (CCC) server, a dedicated Dell Inspiron 530 desktop,
automated the bulk of tasks. Programs written in the MATLAB language
are scheduled to download real-time data, assess data quality, perform
calculations, store/archive data, and update web pages. Real-time data
sources include pressure and temperature data from RTWOS, winds at
station DISW3 from the National Data Buoy Center (Fig. 1), and radarre-
ectivity images at station DLH from the NWS. Contingency data
sources downloaded in the event of missing real-time information
(see below) are water surface temperatures from the Great Lakes Sur-
face Environmental Analysis (
glsea.html) and gridded forecast winds (
res/glcfs/). Processing of the gathered data involves multiple tasks.
Wave properties are calculated from measured subsurface pressure
(Tsai et al., 2005;Jones and Monismith, 2007), and radar images are
processed to assess the storm motions. Lastly, the computed SI and
processed data are formatted into a webpage that is uploaded to a
web server.
Second, a High Performance Computing (HPC) server, a workstation
with four Intel Core i7-3930K hex-core processors, is dedicated to exe-
cute time-consuming surface wave and hydrodynamic modeling. The
HPC architecture with 24 cores follows the INFOS-Yahara computing in-
frastructure (Reimer and Wu, 2016). For the Apostle Islands in Lake Su-
perior, the third-generation spectral wave model Simulating WAves
Nearshore (SWAN) is employed with spatial resolution up to 100 m to
resolve the nearshore effects of the APIS. For the detailed description
of the SWAN model settings, resolution, and calibration, readers are re-
ferred to Anderson et al. (2015). Real-time and forecast model winds,
downloaded from the GLCFS, are used to drive the model. Automated
scripts execute the model every 6 h to update real-time conditions
and perform a 60 hour model forecast.
To provide kayakers with direct access to real-time water environ-
mental conditions and SI rating before launching into the water, an in-
novative solar powered display, Real-Time Wave Kiosk System
(RTWKS), is designed and implemented inside a weather shelter at
Meyers Beach access where NPS staff is typically posted (Fig. 4a).
RTWKS consists of a communication unit, a low power controlling com-
puter, and a monitor inside a weatherproof enclosure with anti-glare
screen (Fig. 4b). Specically, a Raven X modem by Sierra Wireless
Fig. 3. (a) Schematic and electronic components of RTWOS. (b) Field installation of RTWOS components at the cliff top at the Mainland Sea Caves.
994 J.D. Anderson, C.H. Wu / Journal of Great Lakes Research 44 (2018) 9901001
with an additional Rx diversity antenna to improve the low cellular sig-
nal quality in the remote area is used. A Raspberry Pi 2 computer is
auto-scheduled to run a Linux shell script that refreshes the web page
( displayed on the 22"
LCD display monitor. Due to power demand of the monitor (~30 W)
and the absence of landline power supply, power consumption of
each component is considered and designed to sustain the RTWKS
using solar energy.
Design of the power efcient RTWKS is based upon achieving a bal-
ance between power consumption of the system and power charging
capability of a solar panel size. To limit power consumption of the mon-
itor, a two minute delay timer switch is connected in series with the
monitor so power is only drawn when users activate the kiosk with
the red push button. Upon activation, power is supplied to the monitor
and delay impulse circuit, which provides a pulse of current to the sole-
noid that presses the power button on the display. We employ the
CIEMAT model (Copetti et al., 1993) for battery state-of-charge (SOC)
to optimally size the lead acid battery capacity a nd solar array. Efciency
of the chargingsystem is conservatively estimated at 0.62 (Marion et al.,
2005), and a seven day absence of sunlight due to cloudy weather is as-
sumed. Fig. 5 shows the modeled time series of battery SOC for the
selected 160 Ah battery capacity and 40 watt solar array. The crucial en-
ergy savings from the time delay switch is evidenced by the ~6 days to
drain battery SOC for the 0 minute time interval, i.e. monitor always
on. For more reasonable activation intervals tested, the 160 Ah battery
capacity is necessary to maintain the SOC above a healthy ~0.6 for the
assumed 7 day absence of solar input due to weather. Balance between
solar power input and output occurs at a 17.3 minute activation interval,
which is nearly equal to the 20 minute display design estimate for
activation interval (based upon the information provided by NPS
The following subsections examine a series of observed hazardous
conditions with the computed SI ratings related to kayaker safety.
Moving storm events
Sudden appearance of convective storms is one of the most danger-
ous situations that kayakers can encounter. Multiple recent rescues and
fatalities have occurred in the APIS from several unexpected events [Du-
luth News Tribune, 7/12/2016; 10/07/2015;,09/9/
2016]. In additionto intense convective storms, winds andwave heights
during a storm event can increase rapidly before a kayaker can react.
Fig. 6ab shows time series plots of the forecasted winds and wave
heights (green lines) along with measured winds and wave heights
(blue lines) for a nine day windowin June 2016. It is shown that the syn-
optic scale meteorology is well forecasted on daily scale. Nevertheless,
hourly storm scale forecasts cannot capture two separate convective
storm events that caused wind and wave conditions to spike over the
course of an hour on June 25th (06/25), consistent with the results by
the current state of weather forecast technology (Alves et al., 2014;
Cordi, 2017). As a result, the SI values based upon forecasted wind
and wave conditions, i.e. 57 and 56 for the rst and second event, re-
spectively (green line in Fig. 6d) are higher than the SI values based
upon the measured wind and wave conditions, i.e. 32 and 29 for the
rst and second event, respectively (blue line in Fig. 6d). In short, the
current forecast wind and wave models are capable of predicting fast
moving storm events. SI values based upon measured winds and wave
heights should be used to provide reliable hourly navigational safety as-
sessments for kayakers at the study site.
Radar reectivity images offer an opportunity to address the short-
fall in forecasting hourly convective storms. Fig. 6c shows point totals al-
located to radar in the computation of SI and Fig. 6d shows the SI (red
Fig. 4. (a) RTWKSin a weather shelter at the entrance of Meyers Beach and (b) schematic of a communication unit, a controlling electronic computer, and am anti-glare screen housing.
Fig. 5. Time series of battery state-of-charge (SOC) obtained by the CIEMAT model for
different intervals of RTWKS display activations.
995J.D. Anderson, C.H. Wu / Journal of Great Lakes Research 44 (2018) 9901001
line) with consideration of radar reectivity.SI values with radar are de-
creased approximately 34 h before elevated wind and wave conditions.
For example, before the rst storm event on 6/25, the SI with radar de-
creases to approximately 41. The SI drops below 12 during the highest in-
tensity of the storm. Similar results can be seen for the storm event at the
end of the day on 6/19. Other events from the dataset have been exam-
ined with a similar result, but are not shown for brevity. Future effort
should be paid to address one potential issue that radar reectivity
related to rainfall may not always correlate well with wind speed. As a re-
sult, there are instances when radar causes an underprediction of SI. For
example, on 6/23 the SI with radar dipped to 34 while the SI without
radar remained above 50. Nevertheless, rainfall for a moving storm can
also be a concern for kayaker safety. In short, radar reectivity is an im-
portant component in the automated computation of SI that forewarns
kayakers of a potentially rapidly changing water environment caused
by moving storm events.
Fig. 6. Time series plots from June 2016 of (a) measured (blue) and forecasted (green) wind speedand (b) measured (blue) and forecasted (green) SWH. Point contributions to SI from
wind speed and SWH are shown on the right axis and scale linearly. (c) Time series of spatially averaged radar reectivity (blue) and points allocated to SI (red). (d) Time series of
recommended kayaker class using measured data without radar (blue), forecast data without radar (green), and measured data with radar (red).
Fig. 7. Timeseries in August 2016of (a) signicantwave height (SWH)and SS point contribution on the right axis,(b) spectral steepness (SS) (blue, leftaxis) and SS point contributionto SI
(red, right axis), (c) spectral peakedness parameter Q
(blue, left axis) andQ
point contribution to SI (red, right axis), and (d) kayaking class (left axis) and SI (right axis) with (red) and
without (blue) considering SS and Q
996 J.D. Anderson, C.H. Wu / Journal of Great Lakes Research 44 (2018) 9901001
Extreme and freak wave events
Extreme (or steep) waves at seas can be hazardous to kayakers (Lull,
2013). Fig. 7 shows a representative time series in August 2016 that il-
lustrates wave conditions with allocated points to SI values. Fig. 7a
shows the magnitude of SWH positively determines allocated points
in SI. In this study, we add spectral steepness (SS) in Fig. 7bandspectral
peakedness parameter (Q
)inFig. 7c with both their allocated points in
SI on the right axes. The computed SIs with or without SS and Q
shown in Fig. 7d. For example, the largest waves occurred on 8/21,
yielding the SI = 46 (Class III) with SS and Q
; and SI = 64 (Class II)
without SS and Q
. Note that the SWH at this time was approximately
1 m and among the largest waves at the site (Anderson et al., 2015).
The deduction of approximately 20 points due to SS and QP from the
SI can change one class of recommended skill level, which can be unex-
pected for unexperienced kayakers to rate the extreme waves appropri-
ately for navigation safety.
Freak waves often occur during moderate SWHs with certain spec-
tral properties. Previous studies (Wu and Nepf, 2002;Wu and Yao,
2004) suggest that large SS and Q
values have been associated with in-
creased likelihoodof freak wave occurrence. One possible cause of rapid
variations in SS and Q
values between events is the dynamic nature of
winds, e.g. gustiness. In Fig. 7, it is seen that SS and Q
are generally ob-
served to reach peak values before the peak of SWH, consistent with the
typical evolution of the wind wave spectrum (Sorensen, 2006). Never-
theless, on 08/13 and 08/22 the SWHs were similar, but thecorrespond-
ing SSs were so different, yielding a 12-point difference in SI (Fig. 7d).
Similarly, the two moderate wave events on 08/11 and 08/17 had sim-
ilar SWHs, but the SS and Q
were quite different, yielding a 15-point
difference in SI (Fig. 7d). As a result, SI values with the same SWH
with or without considering SS or Q
can result in one class difference
in recommended skill level, which is likely unexpected to inexperienced
kayakers. In short, bothSS and Q
at the Sea Caves are shown tovary in a
transient and complex manner that can create unexpectedly dangerous
extreme and freak wave conditions. The SI rating system considers the
spectral properties of SS and Q
as a substitute for the breaking waves
category in the original SCRS to reect appropriate class of recom-
mended skill level for kayakers.
Upwelling events
Upwelling at the study site is examined using available data of four
stratied seasons (JulySeptember) from 2014 to 2017. It is found that
23 upwelling events occurred every year with durations ranging
from 2.0 to 12.8 days and a median of 5.2 days, which are consistent
with synoptic scale weather patterns. Observed upwelling events gen-
erally occurred when winds shifted to an easterly direction. Since the
study site is located along a southern shore with an east-west orienta-
tion (Fig. 1), the upwelling events likely occur from Ekman transport
wherein the Coriolis force turns easterly wind generated surface cur-
rents offshore to the north (Rao and Schwab, 2007;Rao and Murthy,
2001). As a result, cold water at depth is upwelled to the surface to re-
place water moving offshore. In the following, we discuss the effects
of coastal upwelling events on changes to water temperature and SI
from two examples. First, Fig. 8 shows a time series of measured
water temperature (WT) at the study site from August. Two upwelling
events during 08/0408/12 and 08/1808/21 occurred with approxi-
mate durations of nine and four days, respectively (Fig. 8a), when
winds increased and shifted to an easterly direction (Fig. 8c and d).
Temperature dropped from 15.7 °C to 7.2 °C in 1.1 days for the rst
event and from 18.8 °C to 8 °C in 0.9 days for the second event. The re-
duced WT lowered SI values at least 1520 points during the upwelling
events (Fig. 8b), which results in a categorical change in recommended
kayaker skill level, e.g. from beginner (I) to intermediate (II) or ad-
vanced (III). For example, on 08/04 a moderate wind speed of 6.6 m/s
and SWH of 0.4 m combined with the cool water temperatures from
the upwelling event to result in an SI of 43 (III), vastly different from
conditions on the previous day (08/03) when SI was as high as 74 (I).
A similar SI drop of 20 points due to an observed upwelling event is
also seen on 08/18. Second, for a typical kayak trip at the study site,
the observed maximum and median of SI change to SI during a three-
hour window at the onset of an upwelling event caused by decreasing
water temperatures is 6.9 and 4.4, respectively. In other words, the
time rate of change of temperatures during the onset of an upwelling
event are typically slow enough such that a SI rating is valid during a
3-hour trip. However, over the course of an upwelling event tempera-
ture decreases can lower total SI to change the kayaker skill level
Fig. 8. Time series in August2014 of (a) water temperature (WT)(blue, left axis) and pointcontributionto SI (red, right axis),(b) kayaking class andtotal SI (right axis),(c) wind speed and
point contribution to SI, and (d) wind direction.
997J.D. Anderson, C.H. Wu / Journal of Great Lakes Research 44 (2018) 9901001
category. The proposed SI method can successfully communicate the
dangers of unexpected upwelling events to kayakers.
Wave direction
Wave direction is known to create unexpected wave conditions due
to sheltering and reection effects. Fig. 9 shows a spatial map of SI at the
Sea Caves for two common wind directions i.e. northwestand northeast
with a constant speed of 10 m/s. To compute SI, spatial variations of
modeled SWH, wave direction, and swim distance from the beach are
accounted while other variables are assumed constant. For the case of
NE waves (Fig. 9a), the SI is N70 (skill class I) in a narrow pathway
that has a low swim distance to safety and smaller wave heights due
to sheltering from the orientation of natural coastline. This area coin-
cides with the kayaking route from the Meyers Beach access point to
the Sea Caves.Nevertheless, conditions quickly becomemore hazardous
from SI = 60 to SI = 40 (skill class II to III) as kayakers are away from
the sheltered pathway. For the case of NW waves (Fig. 9b), there is no
sheltering so that wave heights at Meyers Beach are larger and only a
narrow band of SI = 60 (class II) exists due to low swim distance to
safety. NW waves propagating towards the sea caves can interact with
reected waves to generate larger, steeper, and potentially clapotis
(Fig. 2) waves that can pose a high hazardous condition to kayakers.
As a result, a 10 point reduction for wave direction yields a SI below
30 (skill class IV) in this case. Overall, effects of wave sheltering and re-
ection have potential to create unexpected hazardous conditions that
are considered in the new SI rating system.
Real-time display of SI
Real-time information including multiple environmental variables
and the composite SI is displayed on the RTWKS (Fig. 10). The display
page can also be accessed with the URL
rtwks.html. The SI score is displayed in the center with guidance relat-
ing conditions to recommended skill level by kayaking experts in a
table in the top right (Soares and Powers, 1999;Lull, 2013). Three
hours of past measured wave heights, wind speeds, and water temper-
atures are provided in tabular format. At the bottom left, the computed
SI based upon model results are shown with wave directions to indicate
the potential sheltering or reection effects at the Mainland Sea Caves.
The latest webcam image at the Sea Caves and a radar reectivity ani-
mation are displayed along the remainder of the bottom.While multiple
environmental data and images are presented, a single SI numeric score
with a recommended skill level is prominently displayed. Additional
instructions and explanations are provided to interpret the SI. As a
result, one can effectively grasp or obtain the degree of risk in kayak-
ing and personally examine the environmental factors to make their
own decision if needed. Since 2014, the real-time display SI by
RTWKS has fullled data visualization demands by kayakers, recrea-
tional visitors, and park managers; and successfully sustained to aid
in kayaker safety.
Safety Index rating
Two aspects relatedto SI are discussed here. First, from the aspect of
validity, the devised rating of the Safety Index (SI) is based upon linear
rescaling of the original SCRS using the three categories of environmen-
tal factors. Note that the SCRS was originally designed to rate ocean kay-
aking, which can be different from kayaking in the Great Lakes. For
example, longer wave periods (wave lengths), tides, swells in the
ocean are different from shorter wave periods (wave lengths), seiches,
and moving storm induced meteotsunamis (Anderson et al., 2015;
Fig. 9. Spatial distribution of computedSI for cases when wavesapproach from the (a) northeast (NE) and (b) northwest (NW) directions.Webcam images of waves approaching from the
(c) NE and (d) NW directions.
998 J.D. Anderson, C.H. Wu / Journal of Great Lakes Research 44 (2018) 9901001
Bechle et al., 2016). Here, weightings applied to the different variables
and point allocations to compute SI are based upon the original guide-
lines (Soares and Powers, 1999;Lull, 2013) and recommendations pro-
vided by kayaker experts in the Great Lakes. For new variables in the
category like radar reectivity, wave spectral characteristics, and wave
direction, pointshave been allocated to SI based upon available observa-
tions and INFOS modeling results. In general, it is found that the effects
of new variables can change the recommended kayaker skill level up to
one class. Throughout the four-year testing period (20142017), we
have examined computed SIs with the point allocation in the three cat-
egories in Table 1 for various hazardous conditions of the water envi-
ronment at the Sea Caves in the APIS. Importantly, because of the
cognitive nature of a human safety rating, there is a need to consult
with local kayaking experts to further verify or modify the weightings
of each variable to tune the system in-line with the common users, es-
pecially kayaking experts. Second, from the aspect of adaptability, the
modular design of the SI algorithm makes it easily change when consid-
ering modications to point allocations in specic categories or entirely
new categories.
The SI rating system can be exibly incorporated into a real-time
water environmental cyberinfrastructure system and applied to another
kayaking site where local conditions may differ from the site in this
study. Transferring SI rating system to any new site should be evaluated
for the validity of the categories and weightings considered here and
adapted to differentwater environmental factors. Overall, from both va-
lidity and adaptability aspects, it is shown that the Safety Index rating
can take multiple environmental variables into account to yield one
number to deliver effective real-time risk assessment and communica-
tion of kayaker safety.
Data availability and contingency plan
Data availability is essential to any real-time safety rating system.
Computation of the SI requires a constant stream of real-time data
from multiple sources that may experience downtime. Based on histor-
ical databases, some variables collected by the NWS and NOAA show
reliability of data streams of around 98% availability. The RTWOS has
provided wave and water temperature measurements via cellular net-
work with an availability of 96% over the past four years (20142017).
While each variable has availability that nearly meets federal standards
(National Oceanic and Atmospheric Administration (NOAA), 2016), var-
iables from different systems may not experience similar downtimes.
For example, the net effective availability for all data sources used in
the computation of SI was 93.8% during the 2016 kayak season. Because
multiple data are compiled to compute SI, there is a great need to have a
robust data contingency plan.
Contingency reporting of the SI depends on the source of missing
data. In most cases, one of the three data streams is unavailable, i.e.
RADAR, NOAA meteorological station, or RTWOS. When RADAR is
unavailable, the SI is reported as a range by assuming a best and
worst case scenario for storm conditions, e.g. SI = 5070. Absent
NOAA wind measurements are replaced by gridded winds from the
NWS atmospheric model ±4.2 m/s, which is the historical difference
between model and measured winds at the 95% condence level.
Missing RTWOS data is the most complicated contingency because
it reports wave eld properties and temperature. In the absence of
measured wave information, the SWH is replaced with the wave
model ±0.22 m corresponding to the historical difference between
model and measured SWH at the 95% condence level. Spectral
wave properties are accounted for by applying a range of 20 to the
reported SI but only if SWH is N0.4 m. Inclusion of radar data to the
moving storm events (Fig. 6). Missing temperature data is replaced
with the GLSEA satellite estimated surface temperatures near the
site (Schwab et al., 1999). Examination of the historical dataset has
revealed the GLSEA satellite matches measured temperature values
well but does not resolve coastal upwelling events at the site. There-
the computation of the SI when temperatures are absent. In sum-
mary, the contingency plan to handle missing data has been imple-
mented to the real-time water cyberinfrastructure to compute SI
with a range of the uncertainty.
Fig. 10.An example display ofthe RTWKS that consistof 3 h of past measuredwave heights, windspeeds, and water temperatures,a computed SI witha recommended skilllevel, a SI map
with the consideration of wave direction, a latest webcam image at the Sea Caves, and a radar reectivity animation.
999J.D. Anderson, C.H. Wu / Journal of Great Lakes Research 44 (2018) 9901001
Recreational usage statistics
Usage statistics of the RTWOS website (http://www.seacaveswatch.
org) reveal the public desire for real-time information about the water
environment at the sea caves of the APIS. Website analytics starting
from 2014 shows an average of 83,000 page views annually. Site trafc
tends to follow the kayaking season with webpage visit increasing in
June and July, peaking in August, and decreasing through September
and October. Interestingly, site activity spikes in late January andFebru-
ary to near summer levels as the public and NPS managers examine the
webcam for ice conditions of the so-called winter ice caves. Approxi-
mately 40% of daily trafc in summer months occurs in the morning be-
fore 9 a.m. as the public and kayak outtters check conditions for day
trips. Average activity is relatively constant throughout the remainder
of the day but often spikes during energetic wave events, suggesting
the system is effectively reaching the public with the potential dangers
at the sea cave site. Of the page visits, approximately 40% are from mo-
bile devices but few visits originate from the area of the sea caves. This
suggests that conditions are most often checked before visitors travel to
the sea caves but not once on site, possibly due to poorcellular reception
in the area. Overall, the web analytics demonstrates a public interest in
real-time water conditions of the Sea Caves and the need for the on-site
RTWKS to provide the latest information before kayakers enter the
Summary and conclusions
A water environment cyberinfrastructure has been developed and
applied to provide real-time safety assessment of sea conditions for
kayakers visiting the Mainland Sea Caves of the APIS. The
cyberinfrastructure includes the RTWOS for data acquisition (Fig. 3),
an Integrated Nowcast and Forecast Operation System (INFOS) for
data storage, management, integration, and high performance comput-
ing for weather and wave conditions, and the RTWKS for data visualiza-
tion and display (Fig. 4). A Safety Index (SI) is devised to be a point-
based skill level by considering multiple environmental factors
(Table 1) that characterize navigational safety for kayakers. The SI is
based upon a linear rescaling of the original SCRS, developed by kayak-
ing experts Soares and Powers (1999), and adds new variables to ad-
dress fast moving storms, unique wave characteristics, and rapidly
changing winds in the Great Lakes. Specically, analysis of radar reec-
tivity has been added to improve forecasts of dangerous conditions gen-
erated by convective storms over state-of-the-art weather and wave
modeling by INFOS (Fig. 6). Breaking wave in the original SCRS variable
is replaced by spectral characteristics of surface waves for probabilistic
assessment of extreme and freak waves (Fig. 7). Other unexpectedly
dangerous conditions like coastal upwelling and freak wave occurrence
due to rapidly changing wind directions have been addressed to com-
pute the real-time SI (Figs. 8 and 9). Display of the SI and visualization
of other real-time environmental data are communicated to kayakers,
visitors, or park managers by a power efcient kiosk, RTWKS (Fig. 10),
that is installed at the main access point, Meyers Beach. Designs of
RTWKS and RTWOS for automated eld data acquisition have been
thoroughly tested with attention to sustainable power at an off-grid
site for 4 years. To handle the issue of possibly missing required envi-
ronmental data, a contingency plan using other data sources has been
implemented to the cyberinfrastructure to compute the real-time SI
with a range of the uncertainty. Web analytics demonstrates a public in-
terest in real-time water conditions of the sea caves and the need for the
on-site kiosk to provide the latest information before kayakers enter the
water. The new real-time water environment cyberinfrastructure for
kayaker safety in the Apostle Islands, Lake Superior has been success-
fully operated since 2014. Future work should involve adding a storm
tracking algorithm to the assessment of radar and continuous rene-
ment of SI rating through user feedback.
The authors would like to thank Mr. Gene Clark, a coastal engineer-
ing specialist at the University of Wisconsin Sea Grant Institute (UW-
Sea Grant), for providing valuable knowledge related to the wave im-
pacts on kayaker safety in Lake Superior. We acknowledge Ms. Marie
Zhuikov at UW-Sea Grant for her dedication to communicating coastal
hazards like freak waves and meteotsunami like waves to the Great
Lakes community. We also acknowledge that kayaker experts from In-
land Sea Society, Living Adventures Inc., Lost Creek Adventures, and
Trek & Trail to provide their opinions and suggestions to develop the
proposed Safety Index (SI). The authors specically thank Ms. Julie
Van Stappen, Mr. Bob Krumenaker, Ms. TamHofman, Mr. David Wilkins,
Mr. Damon Panek, and many staff and park rangers at the Apostle
Islands National Lakeshore for their assistance with operation of
RTWOS and RTWKS. We thank Mr. Michael Friis, Mr. Todd Brieby, Ms.
Angel Kathleen, and Mr. Travis Olson at the Wisconsin Coastal Manage-
ment Program (WCMP) for their continuous support on the project
throughout many years. Last but the least, we thank Mr. Larry MacDon-
ald, former Mayor of City of Bayeld, WI, for his vision to facilitate real-
time water environment cyberinfrastructure. This research was partly
supported by the University of Wisconsin Sea Grant Institute under
the grant award NA10OAR4170070, WCMP grants under the grant
awards AD089091-009.23, ADA000213, AD129611-013.25, and
AD169127-017.21, and gift funds by Friends of the Apostle Islands Na-
tional Lakeshore.
Aadland, E., Noer, G., Vikene, O.L., 2016. Sea kayaking incidents in Norway 20002014: an
Alves, J., Chawla, A., Tolman, H., Schwab, D., Lang, G., Mann, G., 2014. The operational im-
plementation of a Great Lakes wave forecas ting system at NOAA/NCEP. Weather
Forecast. 29.
Anderson,J.D., Wu, C.H., Schwab, D.J., 2015.Wave climatology in the Apostle Islands, Lake
Superior.J.Geophys.Res.Oceans120:48694890. https:/ / 1002/
Bailey, I., 2010. An analysis of sea kayaking incidents in New Zealand 19922005. Wilder-
ness Environ. Med. 21 (3):208218.
Bechle, A.J., Wu, C.H.,Kristovich, D.A.R., Anderson, E.J., Schwab, D.J., Rabinovich, A.B., 2016.
Meteotsunamis in the Laurentian Great Lakes. Sci. Rep. 6.
Bishop, C., Donelan, M., 1987. Measuring waves with pressure transducers. Coast. Eng. 11
(4), 309328.
Bogrien, D.W., Brooks, A.S., 1992. Analysis of thermal features of Lake Michi gan from
AVHRR satellite images. J. Great Lakes Res. 18, 259266.
Chen, C.S., Xu, Q., Ralph, E., Budd, J.W., Lin, H., 2004. Response of Lake Superior to mesoscale
wind forcing: a comparison between currents driven by QuikSCAT and buoy winds.
J. Geophys. Res. Oceans 109 (C10):C10S02.
Copetti, J.B., Lorenzo, E., Chenlo, F., 1993. A general battery model for PV system simula-
tion. Prog. Photovolt. Res. Appl. 1, 283292.
Cordi, S., 2017. Forecasting severe convective storms. Oxford Research Encyclopedia of
Climate Science
Csanady, G.T., 1984. Circulation in the Coastal Ocean. D. Reidel Publishing Co., Boston.
Dahle, E.A., Myrhaug, D., 1995. Risk analysis applied to capsize of shing vessels. Mar.
Technol. 32, 245257.
Dean, R.G., 1990. Freak waves: a possible explanation. In: Torum, O., Gudmestad, O.T.
(Eds.), Water Wave Kinematics. Kluwer Acad., Norwell, Mass, pp. 609612.
Dysthe, K., Krogstad, H.E., Muller, P., 2008. Oceanic rogue waves. Annu. Rev. Fluid Mech.
Geist, E.L., ten Brink, U.S., Gove, M., 2014. A framework for the probabilistic analysis of
meteotsunamis. N at. Hazards 74 (1):123142.
Goda, Y., 1970. Numerical experiments on wave statistics with spectral simulation. Rep.
Port Harb. Res. Inst. 9 (3).
Hsu, S.A.,Meindl, E.A., Gilhousen, D.B.,1994. Determining thepower-law wind-prole ex-
ponent under near-neutral stability conditions at sea. J. Appl. Meteorol. 33 (6 ):
Hultquist, T., Dutter, M., Schwab, D., 2006. Reexamination of the 910 November 1975
Edmund Fitzgeraldstorm using today's technology. Am. Meteorol. Soc. 87 (5),
Johns, R.H., Hirt, W.D., 1987. Derechos: widespread convectively induced windstorms.
Weather Forecast. 2:3249. .1175/1520-0434(19 87)002b0032:
Jones, N.L., Monismith, S.G., 2007. Measuring short-period wind waves in a tidally forced
environment with a subsurface pre ssure gauge. Limn ol. Oceanogr. Meth ods 5,
1000 J.D. Anderson, C.H. Wu / Journal of Great Lakes Research 44 (2018) 9901001
Kharif, C., Pelinovsky, E., 2003. Physicalmechanisms of the rogue wave phenomenon. Eur.
J. Mech. B. Fluids 22, 603634.
Kraft, G.J., Mechenich, C., Mechenich, D.J., Szczytko, S.W., 2007. Assessment of Coastal
Water Resources and Watershed Conditions at Apostle Islands National Lakeshore,
Wisconsin. Natural Resource Technical Report NPS/NRWRD/NRTR2007/367. Na-
tional Park Service, Fort Collins, Colorado Available at.
Liu, P.C., Schwab, D.J., Bennett, J.R., 1984. Comparison of a two-dimensional wave predic-
tion model with synopticmeasurements in Lake Michigan. J. Phys. Oceanogr. 14 (9):
Liu, P.C., Wu, C.H., Bechle, A.J., MacHutchon, K.R., Chen, H.S., 2010. What do we know
about freaque waves in the ocean and lakes and how do we know it. Nat. Hazards
Earth Syst. Sci. 10, 21912196.
Lull, J., 2013. Sea Kayaking Safety and Rescue. Wilderness Press, Berkeley, CA.
Marion, B., Adels tein, J., Boyle, K., Hayden, H., Hammond, B., Fletcher, T., Canada, B.,
Narang, D., Shugar, D., Wenger, H., Kimber, A., Mitchell, L., Rich, G., Townsend, T.,
2005. Performance parameters for grid-connected PV systems. Proceedings of 31st
IEEE PVSC 2005, Lake Buena Vista, U.S.A., pp. 16011606
Naess, A., 1985. On the distribution of crest to trough wave heights. Ocean Eng. 12 (3),
NationalOceanic and Atmospheric Administration(NOAA), 2016. NEXRADStrategic Plan.
By T. Clark. Available at:.
Nepf, H.M., Wu, C.H., Chan, E.S., 1998. A comparison of two- and three-dimensional wave
breaking. J. Phys. Oceanogr. 28, 14961510.
Rabinovich, A.B., 2009. Seiches and harbor oscillations. In: Kim, Y.C. (Ed.), Handbook of
Coastal and Ocean Engineering. World Scientic Publishing Company, Toh Tuck
Ragotzkie, R.A., 1974. Vertical motions along the north shore of Lake Superior. Proceed-
ings of the 17th Conference of Great Lakes Research, pp. 456461.
Rao, Y.R., Murthy, C.R., 2001. Nearshore currents and turbulent exchange processes dur-
ing upwelling and downweling events in Lake Ontario. J. Geophys. Res. 106 (C2),
Rao, Y.R., Schwab, D.J., 2007. Transport and mixing between the coastal and offshore wa-
ters in the Great Lakes: a review. J. Great Lakes Res. 33, 202218.
Reimer, J.R., Wu, C.H., 2016. Development and application of a nowcast and forecast sys-
tem tool for planning and managing a river chain of lakes. Water Resour. Manag. 30
(4), 13751393.
Schwab, D.J., Bennett, J.R., Lynn, E.W., 1984. A Two-dimensional Lake Wave Prediction
System, TM-051.
Schwab, D.J., Leshikevich, G.A., Muhr, G.C., 1999. Automated mapping of surface water
temperature in the Great Lakes. J. Great Lakes Res. 25 (3), 468481.
Scott, R.W., Huff, F.A., 1996. Impacts of the Great Lakes on regional climate conditions.
J. Great Lakes Res. 22 (4), 845863.
Soares, E., Powers, M., 1999. Extreme Sea Kayaking. Ragged Mountain Press, Camden,
Sorensen, R.M., 2006. Basic Coastal Engineering. Third Edition. Springer, New York.
Stopa, J.E., Filipot, J.-F., Li, N., Cheung, K.F., Chen, Y.-L., Vega, L., 2013. Wave energy re-
sources along the Hawaiian Islands chain. Renew. Energy 55, 305321.
Tew, M.A., Mann, G.E., Jillson, E., Nelaimischkies, S., May, R., 2008. Implementation of a
new Marine Weather Wa rning Bulletin by the National Weather Service. Article
131866. 24th Interactive Information and Processing Systems Conference, 88th
American Meteorological Society, New Orleans, LA.
Tsai, C.H., Huang, M.C., Yound, F.J., Lin, Y.C., Li, H.S., 2005. On the recovery of surface wave
by pressure transfer function. Ocean Eng. 32, 12471259.
Wu, C.H., Nepf, H.M., 2002. Breaking criteria and energy losses for three-dimensional
wave breaking. J. Geophys. Re s. Oceans 107 (C10):3177. ht tps://
Wu, C.H., Yao, A.F., 2004. Laboratory measurements of limiting freak waves on currents.
J. Geophys. Res. Oceans 109 (C12), C12002.
Yao, A., Wu, C.H., 2005. Incipient breaking of unsteady waves on sheared currents. Phys.
Fluids 17, 082104.
1001J.D. Anderson, C.H. Wu / Journal of Great Lakes Research 44 (2018) 9901001
... General description of the INFOS is described here while the details of INFOS cyberinfrastructure can refer to Reimer and Wu (2016) and Anderson and Wu (2018). Fig. 3a shows the three components of computing infrastructure of INFOS: a locally-based High-Performance Computing (HPC) cluster, a cloud-based distributed HPC grid, and a secure web server. ...
... Despite the different viewpoints on water environment safety by different scholars, the core and essence are basically the same. From the environmental aspect, water environment safety means the ability of human beings to obtain the amount of water that meets their basic physiological and living needs [20]. From the economic perspective, water environment safety means that the socioeconomic system is able to obtain the amount of water that enables sustainable development [21]. ...
Full-text available
Focusing on the topic of water environment safety of China, this paper has selected the three northeast provinces of China as the research object due to their representativeness in economic development and resource security. By using the Entropy Weight Method, the Grey Correlation Analysis Method, and the Principal Component Analysis Method, this paper has first constructed a water environment safety evaluation system with 17 indicators from the economic, environmental, and ecological aspects. Furthermore, this paper has screened the initially selected indicators by the Principal Component Analysis Method and finally determined 11 indicators as the evaluation indicators. After indicator screening, this paper has adopted the improved Fuzzy Comprehensive Evaluation Method to evaluate the water environment safety of the three northeast provinces of China and obtained the change in water environment safety of different provinces from 2009 to 2017. The results show that the overall water environment safety of the region had improved first but worsened afterward, and that in terms of water safety level, Jilin Province ranked first, followed by Heilongjiang Province and Liaoning Province. The three factors that have the greatest impact on the water environment safety of the three provinces are: Liaoning—Chemical Oxygen Demand (score: 17.10), Per Capita Disposable Income (score: 13.50), and Secondary Industry Output (score: 11.50); Heilongjiang—Chemical Oxygen Demand (score: 18.64), Per Capita Water Resources (score: 12.75), and Concentration of Inhalable Particles (score: 10.89); Jilin—Per Capita Water Resources (score: 15.75), Chemical Oxygen Demand (score: 14.87), and Service Industry Output (score: 11.55). Based on analysis of the evaluation results, this paper has proposed corresponding policy recommendations to improve the water environment safety and promote sustainable development in the northeast provinces of China.
Full-text available
Rip currents near coastal structures commonly occur in Lake Michigan in the Great Lakes region of the United States. Lack of timely warning due to undocumented characteristics of rip currents and no assessment tool can contribute to tragic drownings incidents. In this paper, we characterized rip current occurrences near breakwater structures and developed an assessment tool for providing timely rip current warnings to beachgoers at the study site, City of Port Washington, WI. Characteristics of rip currents near the structure were observed from field measurements or visual images. Deflection rip currents had speeds of ∼ 0.2 m/s and lasted for several hours. The rip current occurrences were associated with environmental proxies. It was found that rip currents can occur even when the water appears calm near the structure. A Structure Rip Checklist and Assessment Matrix (SRiCAM) with a four-tiered risk was developed and validated using observations. Furthermore, the SRiCAM was integrated into cyberinfrastructure with a data contingency plan to provide real-time warnings to the public. The applicability of the SRiCAM to other locations across Lake Michigan was further tested and results are promising. Overall, the SRiCAM has the potential to be widely extended to foster recreational water safety and resilience to rip current hazards in the Great Lakes.
Full-text available
Forecasting severe convective weather remains one of the most challenging tasks facing operational meteorology today, especially in the mid-latitudes, where severe convective storms occur most frequently and with the greatest impact. The forecast difficulties reflect, in part, the many different atmospheric processes of which severe thunderstorms are a by-product. These processes occur over a wide range of spatial and temporal scales, some of which are poorly understood and/or are inadequately sampled by observational networks. Therefore, anticipating the development and evolution of severe thunderstorms will likely remain an integral part of national and local forecasting efforts well into the future. Modern severe weather forecasting began in the 1940s, primarily employing the pattern recognition approach throughout the 1950s and 1960s. Substantial changes in forecast approaches did not come until much later, however, beginning in the 1980s. By the start of the new millennium, significant advances in the understanding of the physical mechanisms responsible for severe weather enabled forecasts of greater spatial and temporal detail. At the same time, technological advances made available model thermodynamic and wind profiles that supported probabilistic forecasts of severe weather threats. This article provides an updated overview of operational severe local storm forecasting, with emphasis on present-day understanding of the mesoscale processes responsible for severe convective storms, and the application of recent technological developments that have revolutionized some aspects of severe weather forecasting. The presentation, nevertheless, notes that increased understanding and enhanced computer sophistication are not a substitute for careful diagnosis of the current meteorological environment and an ingredients-based approach to anticipating changes in that environment; these techniques remain foundational to successful forecasts of tornadoes, large hail, damaging wind, and flash flooding.
Full-text available
The generation mechanism of meteotsunamis, which are meteorologically induced water waves with spatial/temporal characteristics and behavior similar to seismic tsunamis, is poorly understood. We quantify meteotsunamis in terms of seasonality, causes, and occurrence frequency through the analysis of long-term water level records in the Laurentian Great Lakes. The majority of the observed meteotsunamis happen from late-spring to mid-summer and are associated primarily with convective storms. Meteotsunami events of potentially dangerous magnitude (height > 0.3 m) occur an average of 106 times per year throughout the region. These results reveal that meteotsunamis are much more frequent than follow from historic anecdotal reports. Future climate scenarios over the United States show a likely increase in the number of days favorable to severe convective storm formation over the Great Lakes, particularly in the spring season. This would suggest that the convectively associated meteotsunamis in these regions may experience an increase in occurrence frequency or a temporal shift in occurrence to earlier in the warm season. To date, meteotsunamis in the area of the Great Lakes have been an overlooked hazard.
Full-text available
A nowcast and forecast system for providing real-time water information of a River Chain of Lakes (RCL) is developed. The system infrastructure comprises a web portal to retrieve and display observations that are used to drive models under a high performance computing server. Water level and flow discharge information are obtained from a suite of models that be directly simulate the RCL system. A new data assimilation technique based upon flow routing algorithm and nested-mesh domain reduction is developed to update the Manning’s roughness. It is demonstrated that the INFOS can reliably and effectively model real-time reverse flows due to sustained wind forcings or tranisent seiches, and flow choking due to channel constriction. Applications of the developed system are illustrated. Specifically water level planning scenarios provide a quantitative measure for lake management to reduce floods under extreme rainfall events. Alternative management philosophies to minimize exceeding the water level orders are evaluated. Overall, the Integrated Nowcast and Forecast Operation System (INFOS) provides reliable and timely water information for the RCL for sharing information to the community, planning for water use and delivery, and management of the Yahara RCL.
Full-text available
The aims of this study were to analyse recreational sea kayaking and touring incidents in Norway with a specific focus on wind conditions and to elaborate on practical implications for the prevention of future incidents. We included 49 incidents reported by the media between 2000 and 2014. Incidents occurred in various wind conditions, but most incidents (60%) occurred in moderate to strong breezes (9–14 m/s). Wind strength and direction were generally forecast accurately and conditions were mainly stable throughout the day of the incident. Thus, sea kayaking and touring incidents in Norway seem to happen in various wind conditions; however, paddlers should have been well informed and aware of the hazards they were facing. Future incidents could be prevented by increasing sea kayakers’ situation awareness through discussion of experts’ decision-making processes and the arrangement of situated learning experiences in realistic settings.
Full-text available
The wave climate of the Apostle Islands in Lake Superior for 35 year (1979–2013) was hindcast and examined using a third-generation spectral wave model. Wave measurements within the Apostle Islands and offshore NOAA buoys were used to validate the model. Statistics of the significant wave height, peak wave period, and mean wave direction were computed to reveal the spatial variability of wave properties within the archipelago for average and extreme events. Extreme value analysis was performed to estimate the significant wave height at the 1, 10, and 100 year return periods. Significant wave heights in the interior areas of the islands vary spatially but are approximately half those immediately offshore of the islands. Due to reduced winter ice cover and a clockwise shift in wind direction over the hindcast period, long-term trend analysis indicates an increasing trend of significant wave heights statistics by as much as 2% per year, which is approximately an order of magnitude greater than similar analysis performed in the global ocean for areas unaffected by ice. Two scientific questions related to wave climate are addressed. First, the wave climate change due to the relative role of changing wind fields or ice covers over the past 35 years was revealed. Second, potential bluff erosion affected by the change of wave climate and the trend of lower water levels in the Apostle Islands, Lake Superior was examined.
This chapter presents an overview of seiches and harbor oscillations. Seiches are long-period standing oscillations in an enclosed basin or in a locally isolated part of a basin. They have physical characteristics similar to the vibrations of a guitar string or an elastic membrane. The resonant (eigen) periods of seiches are determined by basin geometry and depth and in natural basins may range from tens of seconds to several hours. The set of seiche eigen frequencies (periods) and associated modal structures are a fundamental property of a particular basin and are independent of the external forcing mechanism. Harbor oscillations (coastal seiches) are a specific type of seiche motion that occur in partially enclosed basins (bays, fjords, inlets, and harbors) that are connected through one or more openings to the sea. In contrast to seiches, which are generated by direct external forcing (e.g., atmospheric pressure, wind, and seismic activity), harbor oscillations are mainly generated by long waves entering through the open boundary (harbor entrance) from the open sea. Energy losses of seiches in enclosed basins are mostly due to dissipative processes, while the decay of harbor oscillations is primarily due to radiation through the mouth of the harbor. An important property of harbor oscillations is the Helmholtz mode (pumping mode), similar to the fundamental tone of an acoustic resonator. This mode is absent in a closed basin. Harbor oscillations can produce damaging surging (or range action) in some ports and harbors yawing and swaying of ships at berth in a harbor. A property of oscillations in harbors is that even relatively small vertical motions (sea level oscillations) can be accompanied by large horizontal motions (harbor currents), resulting in increased risk of damage of moored ships, breaking mooring lines as well as affecting various harbor procedures. Tsunamis constitute another important problem: catastrophic destruction may occur when the frequencies of arriving tsunami waves match the resonant frequencies of the harbor or bay. Seiches, as natural resonant oscillations, are generated by a wide variety of mechanisms, including tsunamis, seismic ground waves, internal ocean waves, and jet-like currents. However, the two most common factors initiating seiches are atmospheric processes and nonlinear interaction of wind waves or swell. At certain places in theWorld Ocean, waves due to atmospheric forcing (atmospheric gravity waves, pressure jumps, frontal passages, squalls) can be responsible for significant, even devastating harbor oscillations, known as meteorological tsunamis. They have the same temporal and spatial scales as typical tsunami waves and can affect coasts in a similar damaging way. © 2018 by World Scientific Publishing Co. Pte. Ltd. All rights reserved.
Capsize contributes only moderately to the total ship loss rate, however it dominates human loss rate contribution for the smaller vessel. This study suggests that a risk analysis method can be used to manage the risk of capsize. First, dangerous wave events are selected. Following this, the frequency of occurrence is calculated, then the vessel's response to the selected wave events is determined. Finally, the probability of occurrence of wave/vessel encounters per year that will cause capsize is calculated. Examples for two U.S. fishing areas are taken as examples to illustrate the approach.
Basic Coastal Engineering, 3rd Edition offers the basics on monochromatic and spectral surface wave mechanics, coastal water level variations, coastal structures and coastal sedimentary processes. It also provides the necessary background from which the reader can pursue a more advanced study of the various theoretical and applied aspects of coastal hydrodynamics and design. This classic text offers senior and beginning post-graduate students in civil and mechanical engineering or the physical and environmental sciences a well-rounded introduction to coastal engineering. Engineers and physical scientists who have not had the opportunity for formal study in coastal engineering, but would like to become familiar with the subject, will also benefit from this timely resource. New material covered in this third edition includes: Material on coastal processes including beach equilibrium profiles, beach profile closure depth, mechanisms causing beach profile change, and the characteristics and design of coastal entrances. Material on the design of stone mound structures including low-crested breakwaters, sensitivity of the Hudson equation for rubble mound structure design, armor stone specification and the economic implications of design wave selection. Material on surface waves including vessel-generated waves, refraction and diffraction of directional wave spectra and design wave selection examples. © 2006 Springer Science+Business Media, Inc. All rights reserved.
A possible explanation is proposed for the occurrence of freak waves, which are defined as waves with larger heights than expected based on the Rayleigh distribution. The suggested cause is due to nonlinearities of superposition of waves which are not accounted for in the Rayleigh distribution. When two waves combine, if their fundamental components add linearly, it can be shown that the combined wave height increases by more than the sum of the fundamental components. The argument does not address the correctness of linear addition of the fundamental components nor does it include energy closure. An example is presented illustrating the concept.