ArticlePDF Available

Using 3D modeling and remote sensing capabilities for a better understanding of spatio-temporal heterogeneities of phytoplankton abundance in large lakes

Authors:
  • Commission Internationale pour la Protection des Eaux du Léman (CIPEL)
  • French Biodiversity Agency & National Research Institute of Science and Technology for Environment and Agriculture , Aix-en-Provence, France

Abstract and Figures

Lake biological parameters show important spatio-temporal heterogeneities. This is why explaining the spatial patchiness of phytoplankton abundance has been a recurrent ecological issue and is an essential prerequisite for objectively assessing, protecting and restoring freshwater ecosystems. The drivers of these heterogeneities can be identified by modeling their dynamics. This approach is useful for theoretical and applied limnology. In this study, a 3D hydrodynamic model of Lake Geneva (France/Switzerland) was created. It is based on the Delft3D suite software and includes the main tributary (Rhône River) and two-dimensional high-resolution meteorological forcing. It provides 3D maps of water temperature and current velocities with a 1 h time step on a 1 km horizontal grid size and with a vertical resolution of 1 m near the surface to 7 m at the bottom of the lake. The dynamics and the drivers of phytoplankton heterogeneities were assessed by combining the outputs of the model and chlorophyll-a concentration (Chl-a) data from MERIS satellite images between 2008 and 2012. Results highlight physical mechanisms responsible for the occurrence of seasonal hot-spots in phytoplankton abundance in the lake. At the beginning of spring, Chl-a heterogeneities are usually caused by an earlier onset of phytoplankton growth in the shallowest and more sheltered areas; spatial differences in the timing of phytoplankton growth can be explained by spatial variability in thermal stratification dynamics. In summer, transient and locally higher phytoplankton abundances are observed in relation to the impact of basin-scale upwelling.
Content may be subject to copyright.
Using 3D modeling and remote sensing capabilities for a better
understanding of spatio-temporal heterogeneities of phytoplankton
abundance in large lakes
Frédéric Soulignac
a,
,Pierre-AlainDanis
b
,DamienBouffard
c
, Vincent Chanudet
d
, Etienne Dambrine
a
,
Yann Guénand
a,e
, Tristan Harmel
f,g
, Bastiaan W. Ibelings
h
,DominiqueTrevisan
a
,
Rob Uittenbogaard
i
, Orlane Anneville
a
a
French National Institute for Agricultural Research (INRA), CARRTEL, Université Savoie Mont Blanc, 75bis avenue de Corzent, 74200 Thonon-les-Bains, France
b
Agence Française pour la Biodiversité, Pôle AFB-Irstea Hydroécologie Plans d'Eau, 3275 route Cézanne, 13182 Aix-en-Provence, France
c
Eawag, Swiss Federal Institute of Aquatic Science and Technology, Surface Waters Research and Management, Kastanienbaum, Switzerland
d
Hydraulic Engineering Center (CIH), Electricité de France (EDF), Savoie Technolac, 73370 Le Bourget-du-Lac, France
e
SEGULA Technologie, 30 allée du lac d'Aiguebelette BP80271, 73375 Le Bourget du Lac, France
f
Sorbonne Université, Centre National de la Recherche Scientique (CNRS), Laboratoire d'Océanographie de Villefranche, Observatoire Océanologique de Villefranche sur Mer, 181
Chemin du Lazaret, 06230 Villefranche sur Mer, France
g
Irstea, UR RECOVER, Pôle AFB-Irstea Hydroécologie des Plans d'eau, 3275 route Cézanne, 13182 Aix-en-Provence, France
h
Department F.-A. Forel for Environmental and Aquatic Sciences (DEFSE), University of Geneva, Boulevard Carl-Vogt 66, 1211 Geneva, Switzerland
i
Deltares, Boussinesqweg 1, 2629, HV, Delft, The Netherlands
abstractarticle info
Article history:
Received 4 October 2017
Accepted 23 April 2018
Available online 18 June 2018
Communicated by Anthony Vodacek
Lake biological parameters show important spatio-temporal heterogeneities. This is why explaining the spatial
patchiness of phytoplankton abundance has been a recurrent ecological issue and is an essential prerequisite
for objectively assessing, protecting and restoring freshwater ecosystems. The drivers of these heterogeneities
can be identied by modeling their dynamics. This approach is useful for theoretical and applied limnology. In
this study, a 3D hydrodynamic modelof Lake Geneva (France/Switzerland) wascreated. It is basedon the Delft3D
suite software and includes the main tributary (Rhône River) and two-dimensional high-resolution meteorolog-
ical forcing. It provides 3D maps of water temperature and current velocities with a 1 h timestep on a 1 km hor-
izontal grid size and with a vertical resolution of 1 m near the surface to 7 m at the bottom of the lake. The
dynamics and the drivers of phytoplankton heterogeneities were assessed by combining the outputs of the
model and chlorophyll-aconcentration (Chl-a)data from MERIS satellite imagesbetween 2008 and 2012. Results
highlightphysical mechanisms responsible for the occurrence ofseasonal hot-spotsin phytoplanktonabundance
in the lake.At the beginning of spring, Chl-aheterogeneities are usually causedby an earlier onset of phytoplank-
ton growthin the shallowest and more sheltered areas;spatial differencesin the timing of phytoplankton growth
can be explained by spatial variability in thermal stratication dynamics. In summer, transient and locally higher
phytoplankton abundances are observed in relation to the impact of basin-scale upwelling.
© 2018 International Association for Great Lakes Research. Published by Elsevier B.V. All rights reserved.
Keywords:
3D modeling
Remote sensing
chlorophyll-a
Spatio-temporal heterogeneity
Lake Geneva
Delft3D
Introduction
In a world where lakes provide multiple services to society while
facing strong anthropogenic pressures, the protection and restoration
of inland water bodies has become a critical political and scientic
issue. For example, the European Water Framework Directive obliges
member countries to take steps to reach a good ecological status for
all their surface waters (European Commission, 2000). To reach these
challenging goals, the scientic and lake management communities
aim to deepen the understanding of lake ecological functioning, using
ongoing lake monitoring efforts (Stewart et al., 2016).
Phytoplankton, at the base of the food chain, controls to a fair extent
the quality of the ecosystem services and is widely used as an indicator
of ecosystem health and sustainability (Xu et al., 2001). However, eval-
uating the abundance and composition of this community in routinely
obtained water samples raises several issues regarding the representa-
tiveness of the collected lake samples since phytoplankton is rarely ho-
mogeneously distributed over a water body (Pelechaty and Owsianny,
2003). This is all the more true in large and/or deep lakes (Leoni et al.,
2014;Viljanen et al., 2009).
Journal of Great Lakes Research 44 (2018) 756764
Corresponding author at: Ecolog ical Engineering Laboratory (ECOL), Federal
Polytechnic School of Lausanne (EPFL), Station 2, 1015 Lausanne, Switzerland.
E-mail address: frederic.soulignac@ep.ch (F. Soulignac).
https://doi.org/10.1016/j.jglr.2018.05.008
0380-1330/© 2018 International Association for Great Lakes Research. Published by Elsevier B.V. All rights reserved.
Contents lists available at ScienceDirect
Journal of Great Lakes Research
journal homepage: www.elsevier.com/locate/jglr
Understanding the mechanismsthat drive plankton spatial distribu-
tion has been a recurring theme in aquatic ecology and has been studied
for several decades (Arhonditsis et al., 2004). Commonly, phytoplank-
ton spatialheterogeneities fall into two categories, vertical and horizon-
tal distributions. Phytoplankton can only develop in the upper part of
the water column where solar energy is high enough to support net
photosynthesis (Reynolds, 1997), leading to the concepts of critical
depth and turbulence (Huisman et al., 1999). Still, important gradients
across the euphotic zone in both the abundance and composition of
phytoplankton are frequently observed in deep lakes (Pomati et al.,
2011). The vertical distribution of chlorophyll-aconcentration (Chl-a)
in deep lakesis often characterized by the presence of a deep Chl-amax-
imum which acts as a local hotspot of primary production. The occur-
rence and the structure of a deep Chl-amaximum is only now
beginning to be properly understood (Leach et al., 2018). Heteroge-
neities along the horizontal axis that have been observed using tra-
ditional multi-site measurements and more recent techniques such
as microwave or optical remote sensing from space remain equally
difcult to explain because of the complexity of the underpinning
processes and mechanisms that are the basis of these heterogene-
ities (Moreno-Ostos et al., 2007). Indeed, the distribution of phyto-
plankton is strongly affected by wind-driven currents and also by
spatial heterogeneity on environmental conditions (Cyr, 2017;
Huang et al., 2014).
Despite the demonstration of lake patchiness in satellite observa-
tions, horizontal variabilityin lake water parameters is often neglected.
Samples used for lake monitoring and water quality assessment are
usually collected at a single station, often the deepest point of the lake,
because this area is generally located in the center of the lake where an-
thropogenic impact and effect are limited. Nevertheless, the recent re-
mote sensing capabilities can facilitate the description and to some
extent, depending on the frequency of suitable images, the study of dy-
namics in lake heterogeneity (Bresciani et al., 2011;Gons et al., 2008;
Kauer et al., 2015;Odermatt et al., 2012;Palmer et al., 2015). In large
lakes, surface horizontal heterogeneities in Chl-aare a frequent phe-
nomenon and also present structures that could change rapidly
(Kiefer et al., 2015). Thus, the existence of such heterogeneities might
raise several issues when studying the impact of local and global forcing
on phytoplankton that requires a better understanding of the un-
derlying mechanisms responsible for those heterogeneities (Chen
et al., 2003). This is not just a matter of academic interest because
the distribution of blooms of harmful cyanobacteria across lakes is
of prime importance for risk assessment and management and af-
fects the use of lakes for drinking water and recreation (Ibelings
et al., 2003).
The physical properties of water govern its response to mechanical
energy inputs, such as wind, river inow and heat exchanges, resulting
in vertical gradients of physical and chemical characteristics, such as
temperature and nutrients, which are major factors controlling phyto-
plankton development. As such these physical processes may promote
spatial heterogeneities in phytoplankton (Reynolds, 1997). Accordingly,
we postulatehere that a good understanding of physical dynamics is es-
sential to explain the spatial heterogeneity observed in phytoplankton
abundance. To test this, we applied a three-dimensional (3D) hydrody-
namic model to Lake Geneva (France/Switzerland) and evaluated the
ability of this hydrodynamic model to explain horizontal patchiness in
phytoplankton abundance. For that purpose, we analyzed the vertical
thermal structure predicted by the model at different stations in the
lake when Chl-apresented strong surface heterogeneities. The role of
hydrodynamic processes was expected tobe strong enough that consid-
ering hydrodynamics alone would be sufcient to explain some of the
phytoplankton heterogeneity patterns observed in Lake Geneva from
2008 to 2012. In other words, the purpose of this study is to show that
by considering only hydrodynamics, we are able to explain and predict
certain surface phytoplankton heterogeneities that are observed in
lakes.
Materials and methods
Study site
Lake Geneva is a large and deep peri-alpine lake located in the west-
ern part of theAlps, on the border between France and Switzerland (Fig.
1and Table 1). According to the international commission for the pro-
tection of Lake Geneva (CIPEL), it is thermally stratied during much
of the year, never freezes over and does not undergo complete mixing
every year. Its main tributary is the River Rhone which with
184.3 m
3
/s, accounts for 85% of the total inow (average from 1965 to
2015). The lake is monitored as part of a long-term in situ monitoring
program by CIPEL for water quality (e.g. water temperature, Chl-a,
transparency) and biological compartments, including the phytoplank-
ton. This monitoring revealed important changes in phosphorus con-
centrations. In fact, Lake Geneva was eutrophic for several years in the
1960s and measures to reduce phosphorus in its watersheds were rst
implemented in the 1970s, leading to a decrease in phosphorus concen-
trations starting in the early 1980s. Today, Lake Geneva is mesotrophic
(Jacquet et al., 2014a).
Data
In this study, in situ vertical proles of Chl-aand water temperature
(WT), transparency and remote sensed surface Chl-amaps are used be-
tween 2008 and 2012. In situ Chl-a, WT and transparency are available
from the CIPEL monitoring program. Sampling takes place at two stations,
at the deepest point of the lake in the large basin (monitoring station
SHL2: WGS84 6.58872° E, 46.45270° N; depth: 309 m) and in the small
basin (monitoring station GE3: WGS84 6.21994° E, 46.29721° N; depth:
72 m) (Fig. 1). At SHL2, sampling is conducted twice a month, except in
winter, when it is carried out only once a month. At GE3, sampling fre-
quency is once a month throughout all the year. Samples for Chl-amea-
surements are collected at 10 depths: 0, 1, 2, 3.5, 5, 7.5, 10, 15, 20 and
30 m. Cells are collected on a Whatman GF/C lter (47 mm) and soni-
cated. The pigments are extracted with 90% (v/v) acetone/water and
the solution is ltered through a glass ber lter GF/C (25 mm). Chl-ais
measured by spectrophotometry (Strickland and Parsons, 1968). WT is
measured using multiparameter probes, SST-CTD009, SST-CTM214 and
RBR-XRX-620. The transparency is measured as Secchi disk depth
(SDD) which corresponds to the depth at which the light intensity in
the water column is 15% of the intensity of the surface (Lemmin, 1995).
Remotely sensed Chl-ais available from Kiefer et al. (2015).Itwasde-
rived from satellite MERIS observations and the processing was done
with the FUB WeW neural network algorithm (Schroeder et al., 2007).
The algorithm gives a resolution of 260 × 290 m and can handle Chl-a
Inflow
S1
S2 (SHL2)
S3
S4 (GE3)
Outflow 10 km
10 km
North
Fig. 1. Lake Geneva contours, isodepth s (100, 200 and 300 m), numerical domain
(curvilinear grids), location of the main tributary inow and outow (Rhône River) and
location of the four sites where simulation outputs are analyzed (S1, S2 (SHL2), S3, S4
(GE3)).
757F. Soulignac et al. / Journal of Great Lakes Research 44 (2018) 756764
concentrations between 0.05 and 50 μg/L. Data availability is summarized
(Fig. 2).
Delft3D model
The open source Delft3D software used in this study has been widely
applied for lakes of different sizes all around the world (Chanudet et al.,
2012;Kacikoc and Beyhan, 2014;Li et al., 2015;McCombs et al., 2014;
Soulignac et al., 2017;Wahl and Peeters, 2014). A previous modeling ef-
fort was performed on Lake Geneva by using Delft3D. It is presented in a
series of three articles (Razmi etal., 2013, 2014 and 2017). We chose the
same surface heat ux module, the same bottom and surface roughness
formula, the same meteorological inputs and the same Rhone River in-
puts. We also simulated Lake Geneva entirely. But whereas Razmi and
coauthors divided each year into four periods of three months, our ob-
jective was to run one-year simulations to cover the whole thermal
stratication period. Also, they used a very ne grid (the lake surface
was meshed using 45,000 grid cells) to study the small-scale hydrody-
namics in the Vidy bay, a small part of the lake located in front of Lau-
sanne, and they used a non-uniform vertical mesh (sigma-model).
However, we wanted a model requiring less computational time, a reg-
ular resolution throughout the lake and a uniform vertical mesh to bet-
ter describe the thermal stratication. For these reasons, we decided to
create a new model of Lake Geneva. Delft3D has several modules (e. g.:
hydrodynamics, water quality, etc.). In this study, only the hydrody-
namic module Delft3D-FLOW was used. It solves the Navier-Stokes
equations for an incompressible uid, under the shallow water and
the Boussinesq equations. The system of equations consists of the conti-
nuity equation,
u
xþv
yþw
z¼0;
the two horizontal equations of motion,
u
tþuu
xþvu
yþwu
z¼
1
ρ0
p
xþ
xνH
u
x

þ
yνH
u
y

þ
zνV
u
z

þfv and
v
tþuv
xþvv
yþwv
z¼
1
ρ0
p
yþ
xνH
v
x

þ
yνH
v
y

þ
zνV
v
z

fu;
the vertical equation of motion being reduced to the hydrostatic pres-
sure equation, and the transport equation of heat,
T
tþuT
xþvT
yþwT
z¼
xDH
T
x

þ
yDH
T
y

þ
zDV
T
z

þS
ρ0cpw
;
where x,y,andzare Cartesian coordinates (m), tis time (s), u,vand w
are the three components of the water velocity (m/s), fis the Coriolis
frequency in 1/s, Tis the water temperature (K), ρ
0
is the water density
(kg/m
3
), pis the pressure (Pa), ν
H
and ν
V
are the horizontal and vertical
eddy viscosities (m
2
/s), D
H
and D
V
are the horizontal and vertical coef-
cients of eddy diffusivity of heat (m
2
/s), Sis the source of heat per unit
volume (W/m
3
)andc
pw
is the water specic heat (J/K/kg). Model equa-
tions are precisely described (Deltares, 2014).
Model set-up
The surface of the numerical domain was created using a curvilinear
grid composed of 591 cells of about 1 km
2
area to t the lake contour
(Fig. 1). In the vertical direction, 100 Z-layers were used to t the ba-
thymetry. Their thickness varies from 1 m at the surface to about 7 m
at the bottom. Going deeper, the thickness of two consecutive layers in-
creases by a factor of 1.02. Based on these length scales, the computa-
tional time step was set to 3 min to verify the Courant-Friedrichs-
Lewy (CFL) criterion and the background horizontal eddy viscosity
and diffusivity was set to 100 m
2
/s. The k-εturbulence closure model
was chosen to calculate the vertical eddy viscosity and diffusivity be-
cause it has proven to perform well in stratied water (Burchard and
Baumert, 1995). The explicit multi-directional upwind (MDUE) scheme
was selected for the spatial discretization of the horizontal advection
terms and the Van Leer-2 scheme was used for the transport equation.
The wind drag coefcient was set as a linear function of wind speed be-
tween 0.63 10
3
at 0 m/s and 7.23 10
3
at 100 m/s. Stanton and Dalton
numbers were set at 1.3 10
3
. The salinity was set constant at 0.15 ppt,
corresponding to the observed specic conductivity of about 300 μS/cm.
The sediment transport and resuspension in littoral zones and shallow
areas was neglected because no measurements are available to validate
this part of the model and also because we did not rene the grid near
the lake shore.
Model input data
Seven meteorological variables (air pressure, air temperature, cloud
coverage, relative humidity, incident solar radiation, wind speed and
wind direction) coming from simulation results of the Consortium for
Small-scale Modeling (COSMO) two-dimensional atmospheric model
were used in this study (Fig. 3). The time step is 1 h and the spatial res-
olution is1.5 km × 1.5 km. Hourly insitu measurements of the discharge
and the water temperature in the Rhône River at Porte De Scex, 5 km
upstream from the lake were also used (Fig. 3).
Simulations
The simulation starts on March 3, 2008, at midnight. A WT prole
measured at SHL2 at this date was used for initialization considering
the whole lake water temperature as horizontally homogeneous. The
lake was also supposed to be at rest, without water currents. Five
years were simulated, from 2008 to 2012. The simulation was stopped
and restarted on January 1 of each year just to change the transparency
value. The average SDD value between May and September, the ve
months when the solar radiation is the highest, was used (Wahl and
Peeters, 2014). Values are presented in Table 2. Simulation results
were exported with a time-resolution of 1 h.
Table 1
Lake Geneva characteristics.
Characteristic Value
Latitude N 46° 27
Longitude E 6° 32
Elevation 372 m
Mean depth 153 m
Max depth 309 m
Surface area 580 km
2
Watershed area 7475 km
2
2008 2009 2010 2011 2012
Surface Chla
GE3
SHL2
Time
Fig. 2. Temporal availability of in situ data at lake stations and satellite imagery.
758 F. Soulignac et al. / Journal of Great Lakes Research 44 (2018) 756764
Model validation
The model performance was evaluated by comparing observed and
simulated WT at SHL2 and GE3 for the ve simulated years. In this
study, we focused on the water layer between the surface and 30 m
depth because this is where net photoautotrophic phytoplankton
growth occurs. Simulated WT proles were systematically exported
when in situ measurements were taken, at approximately midday.
The model validation was performed by calculating the mean absolute
error (MAE), which quanties the error between simulation results
and observations, dened by
MAE ¼1
nX
n
i¼1
Tsim iðÞTob s iðÞ
jj
where T
sim
and T
obs
are respectively simulated and observed WT vectors
whose length is n. One value of MAE was calculated foreach observation
date by considering all the data between the surface and 30 m. One
value of MAE was also computed for each depth and each year by con-
sidering all the data of one year.
Model analysis
In this paper, we have chosen to focus on surface Chl-astructures
that repeat from time to time. The rst structure was a phytoplankton
patch with enhanced densities near the north-western shore of the
lake in summer. We focused on this particular structure which was ob-
served on a MERIS image on September 7, 2009 (Fig. 4). Accordingly, a
rst period, from August 31 to September 9, 2009, starting with low
and homogenous surface Chl-awas analyzed. The second structure de-
picts higher phytoplankton abundance in early spring in a specicpart
of the lake, namely at the entrance of the Rhône River (Kiefer et al.,
2015) and it was observed on an image dated March 21, 2011 (Fig. 4).
Accordingly, a second period, from March 8 to 28, 2011, was analyzed.
A third period of interest, from February 24 to March 29, 2010, was cho-
sen to be used as a counter-example for the 2nd structure. During that
period, environmental parameters allowed a homogenous develop-
ment of phytoplankton as clearly depicted by remote sensing in
March 23, 2010. Changes in the lake thermal structure during these
three periods were analyzed based on hourly simulated WT proles
for three periods of interest at four stations S1, S2, S3 and S4 (S2 and
S4 correspond to SHL2 and GE3, respectively). Starting and ending
dates of the three periods corresponded to in situ sampling dates at
the station SHL2.
Lake Analyzer is a numerical code that allows thecalculation of sev-
eral lake parameters using high frequency data, it was used to calculate
the mixed layer depth based on hourly simulation results (Read et al.,
2011). The mixed temperature differential parameter of Lake Analyzer
was set to 0.01 °C. For the second and the third period of interest, Lake
Analyzer was also used to calculate the buoyancy frequency, again
based on hourly simulation results, which represents the local stability
of the water column, N
2
, expressed in 1/s
2
by N2¼g
ρ
ρ
z.where gis the
gravity (9.81 m/s
2
), ρis the water mass density(kg/m
3
)andzis the ver-
tical Cartesian coordinate (m).
Results
Model validation
The model was validated by comparing observed and simulated
WT between the surface and 30 m depth at SHL2 (Fig. 5a) and at
GE3 (Fig. 5b). Simulated WT were interpolated at observed depths
and times. The 3D hydrodynamic model accurately reproduced the
observations, in particular, the water column warming from the sur-
face and the implementation of the thermal stratication during the
summer. The MAE calculated for each observation date varies in a
comparable way at SHL2 and GE3 ranging from 0.00 to 3.90 °C at
SHL2 and 0.02 to 3.50 °C at GE3 (Fig. 6a). For both sampling stations,
MAE values are smaller during winter and spring than during sum-
mer and autumn when the lake is stratied. The annual MAE calcu-
lated for each depth considering all the observation dates over the
course of the year also varies comparably between SHL2 and GE3
with values comprised between 0.00 and 1.43 °C at SHL2 and be-
tween 0.22 and 1.89 °C at GE3 (Fig. 6bandFig. 6c). The MAE is gen-
erally minimal at the surface and reaches a maximum at a depth
located in the metalimnion near the thermocline, except for 2009
when the maximum value was found at the surface. In fact, the ther-
mocline depth is better reproduced by the model for the year 2009.
Below the metalimnion, the MAE decreases. The annual mean of
the MAE is systematically lower at SHL2 than at GE3. Based on the
960
1000
1040
AP
(hPa)
0
20
40
AT
(°C)
0
50
100
CC
(%)
0
50
100
RH
(%)
0
500
1000
SW
(W/m2)
0
10
20
U10
(m/s)
0
180
360
α10
(°)
0
500
1000
Qin
(m3/s)
2008 2009 2010 2011 2012
0
10
20
Tin
(°C)
Time
Fig. 3. Model input meteorological data at the monitoring station SHL2 (air pressure(AP),
air temperature (AT),cloud coverage (CC),relative humidity(RH), incident solarradiation
(SW), wind speed(U10) and wind direction(α10)) and hourly ow rate (Qin) and water
temperature (Tin) of the Rhône River).
Table 2
Model input data Secchi disk depth (SDD) from
2008 to 2012.
Year SDD (m)
2008 6.1
2009 6.4
2010 6.1
2011 6.0
2012 4.0
2009
31-Aug 01-Sep 07-Sep
2010
01-Mar 05-Mar 08-Mar 13-Mar 23-Mar
2011
08-Mar 11-Mar 21-Mar
Chl-a (µg/l)
0
2
4
6
8
10
Fig. 4. Satellite chlorophyll-aconcentration (Chl-a)data.
759F. Soulignac et al. / Journal of Great Lakes Research 44 (2018) 756764
MAE value, the best performances are achieved for the year 2012
followed by 2009, 2010, 2011 and 2008 (Table 3).
The comparison between measurements and simulations showed
that the model is capable of reproducing the evolution of the lake ther-
mal vertical structure on a smaller time-scale, especially for the three
time-periods dened previously from surface Chl-aobservations. Dur-
ing the rst period from August 31 to September 9, 2009, the model sat-
isfactorily reproduces the cooling of the water surface temperature and
the deepening of the surface mixed layer at SHL2 (Fig. 7a). MAE values
are presented in Table 4. For the period from February 24 to March 29,
2010, the model also correctly reproduces the homogeneous onset of
the thermal stratication (Fig. 7b). Finally, for the period from March
8 to 28, 2011, the model correctly predicted an early start of the strati-
cation on March 28 at SHL2 while the lake was still not stratied on
March 22 at GE3 (Fig. 7c). So, the model presents good results and can
be safely used for the purpose of this study.
Lake thermal structure in summer 2009
From August 31 to September 2, 2009, simulation results predicted a
homogeneous thermalstratication in the lake between the surface and
30 m depth, except at S1 near the main tributary entering the lake
(Fig. 8b). In fact, the mixed layer depth was about 15 m at S1 and
10 m at S2, S3 and S4. On September 3, a wind eventfrom west or south-
westerly directions induced downwelling at S1 and upwelling at S4.
This wind event continued on September 4 and increased, presenting
wind speed values N10 m/s (Fig. 8a). The minimum thickness of the sur-
face mixed layer calculated by themodel is 5.6 m at S4 and its maximum
value is 25.3 m at S1 (Fig. 8c). Simulation results also predicted that
water at 18.5 °C was brought up to the surface at the north-western
shore (Fig. 9). This temperature value corresponded to a depth of
about 15 m where the maximum of Chl-awas observed from in situ
measurements at SHL2 on August 31 and September 9 (Fig. 10). Inter-
estingly, the satellite image also shows high Chl-aconcentration on Sep-
tember 7 along the north-western shore (Fig. 4).
Lake thermal structure in spring 2010
From February 24 to March 11, 2010, simulation results showed that
thermal stratication did not occur between the surface and 30 m
depth because several wind events (N5 m/s) from west or north-west
and north or north-easterly directions prevented an increase in water col-
umn stability (Fig. 11). Conversely, the wind speed was lower between
March 12 and 17 (b5 m/s) and daily stratication was predicted by the
model. In fact, based on a stability criterion (thermal stratication was de-
ned as N
2
N10
5
1/s
2
), thermal stratication started at the same time at
S1, S2, S3 and S4 on March 16 due to the lower wind speed. Until March
23, the thermal stratication developed homogeneously across the four
stations, no wind events were recorded that could have broken the
Observation
0
10
20
30
Simulation
)
Time
2009 2010 2011 2012
0
10
20
30
WT (°C)
5
10
15
20
25
Observation
0
10
20
30
Simulation
)
Time
Depth (m
2008
Depth (m
2010 2011 2012
0
10
20
30
WT (°C)
5
10
15
20
25
a)
b)
2008 2009
Fig. 5. Comparison betweenobserved and simulatedwater temperature (WT) at SHL2 (a)
and GE3 (b) using contour plots.
Time
2009 2010 2011 2012
MAE (°C)
0
2
4
SHL2
GE3
2008
MAE (°C )
012
Depth (m)
0
10
20
30
SHL2
012
0
10
20
30
GE3
2008
2009
2010
2011
2012
012
Depth (m)
0
10
20
30
2008
012
0
10
20
30
2009
MAE (°C)
012
0
10
20
30
2010
012
0
10
20
30
2011
012
0
10
20
30
2012
SHL2
GE3
a)
b)
c)
Fig. 6. a) Timesseries of mean absolute error (MAE) between thesurface and 30 m depth,
b) Vertical proles of annual MAE.
Table 3
Annual meanvalues of the mean absolute error in watertemperature between the surface
and 30 m depth for the two monitoring stations, SHL2 and GE3, from 2008 to 2012.
SHL2 GE3
2008 1.07 1.30
2009 0.79 0.97
2010 0.80 1.29
2011 1.07 1.20
2012 0.72 0.80
WT (°C)
5 10152025
Depth (m)
0
10
20
30
Observation
WT (°C)
5 10152025
0
10
20
30
Simulation
31-Aug-2009 (SHL2)
09-Sep-2009 (SHL2)
WT (°C)
45678
Depth (m)
0
10
20
30
Observation
WT (°C)
45678
0
10
20
30
Simulation 24-Feb-2010 (SHL2)
17-Mar-2010 (SHL2)
23-Mar-2010 (GE3)
29-Mar-2010 (SHL2)
WT (°C)
56789
Depth (m)
0
10
20
30
Observation
WT (°C)
56789
0
10
20
30
Simulation
08-Mar-2011 (SHL2)
22-Mar-2011 (GE3)
28-Mar-2011 (SHL2)
a)
b)
c)
Fig. 7. Comparison betweenobserved and simulated verticalproles of watertemperature
(WT) at the two monitoring stations, SHL2 and GE3, during the three periods analyzed.
760 F. Soulignac et al. / Journal of Great Lakes Research 44 (2018) 756764
stratication at any station. In comparison, the satellite image March 23
showed quite homogenous Chl-aconcentrations at the whole lake scale
(Fig. 4) and the maximum of Chl-awasobservedfrominsitumeasure-
ments at SHL2 only on March 29 just below the surface (Fig. 10).
Lake thermal structure in spring 2011
From March 8 to 12, 2011, simulation results indicated that the lake
was similarly mixed between the surface and 30 m depth at S1, S2, S3
and S4 when the wind speed did not exceed 5 m/s (Fig. 12). Between
March 13 and 16, several strong wind events (N5 m/s) coming from the
east or southeast regularly reduced the water column stability at S1 by
breaking the small thermal stratication that had begun. On March 13,
the buoyancy frequency at S1 passed from 6 10
5
to 3 10
6
1/s
2
and con-
tinued to reach values below 10
5
1/s
2
every day until March 16. Mean-
while, stability increased at S2, S3 and S4 where the wind speed was
weaker allowing the development of thermal stratication. From March
17, onwards, the stability increased at S1 where the buoyancy frequency
returned over 10
5
1/s
2
due to a decrease in wind speed. On March 19, a
strong wind event from north or northeasterly directions (10 m/s) mixed
the water column at S3 and S4, and signicantly decreased the water col-
umn stability at S2. At S4, the buoyancy frequency passed from 2 10
4
to
310
6
1/s
2
. The wind turned to the west, leaving S1 sheltered and S2 less
exposed to the wind compared to S3 and S4. The north or northeast wind
event continued after March 19 and prevented the thermal stratication
restarting at S3 and S4. As a result, the lake thermal structure was not ho-
mogeneous throughout the lake. Only S1 and S2 were stratied and S1
was more stratied than S2. In comparison, the satellite image on
March 21 showed that algal development took place at S1 (Fig. 4)and
the maximum of Chl-awas observed from in situ measurements at
SHL2 only on March 28 just below the surface (Fig. 10).
Discussion
Model performance
At least ve other 3D lake models which include modeling of water
temperature were created using the open source Delft3D software: Lake
Table 4
Mean absolute error values in water temperature between the surfaceand 30 m depth at
the two monitoring stations, SHL2 and GE3, during the three periods analyzed (NA: not
available).
SHL2 GE3
31/08/2009 1.27 NA
09/09/2009 1.49 NA
24/02/2010 0.52 NA
17/03/2010 0.57 NA
23/03/2010 NA 0.53
29/03/2010 0.54 NA
08/03/2011 0.57 NA
20/03/2011 NA 0.18
28/03/2011 0.54 NA
Fig. 8. First period of interest analyzed in summer 2009, a) Times series of wind speed (U10)
and direction (α10), b) Contour plots of simulated water temperature (WT) at the four
stations,S1,S2,S3andS4,andc)Comparisonofmixed layer depths (metaT) across the lake.
00:00 02:00 04:00
06:00 08:00 10:00
12:00 14:00 16:00
18:00 20:00 22:00
Simulated surface WT (°C)
18.5 19 19.5 20 20.5 21
Fig. 9. Simulated surface watertemperature (WT) on September 4, 2009, every 2 h during
the rst period ofinterest analyzed.
Chl-a (µg/l)
0 5 10 15
Depth (m)
0
10
20
30
2009
31-Aug
09-Sep
Chl-a (µg/l)
0510
0
10
20
30
2010
24-Feb
29-Mar
Chl-a (µg/l)
01020
0
10
20
30
2011
08-Mar
28-Mar
Fig. 10. In situ Chl-ameasurements at SHL2 monitoring station.
761F. Soulignac et al. / Journal of Great Lakes Research 44 (2018) 756764
Geneva in France/Switzerland (Razmi et al., 2017, 2014, 2013), Lake Con-
stance in Switzerland/Germany/Austria (Wahl and Peeters, 2014), Lake
Egirgir in Turkey (Kacikoc and Beyhan, 2014), Nam Theun 2 reservoir in
Laos (Chanudet et al., 2012) and Lake Créteil in France (Soulignac et al.,
2017). All these ve models along with our proposed model, were vali-
dated using in situ measurements and proved to perform satisfactorily,
were capable of accurately simulating the hydrodynamic and thermal
structure of these lakes and proved to be ready to be used for research
questions, for example climate research (Wahl and Peeters, 2014).
Razmi et al.'s model and ours were validated usingvertical proles of
observed water temperature. They used two proles at a single station
in Vidy bay on August 5 and November 7, 2005, while we used
monthly/bimonthly proles at two stations, SHL2 and GE3, in the lake
from 2008 to 2012. Model performances are similar during the studied
periods. They found a RMSE from 0.90 to 2.00 °C and we obtained a MAE
between 0.18 and 1.49 °C during spring and summer.
Early spring algal development onset in 2010 and 2011
In spring 2010, satellite surface Chl-adata showed that the onset of
phytoplankton development was homogeneous in Lake Geneva. This is
corroborated by the use of the 3D model which shows that the thermal
structure of the lake was homogeneous too. By contrast in spring 2011,
satellite images revealed heterogeneous Chl-awith a maximum re-
corded in the eastern part of the lake indicating that the onset of phyto-
plankton development occurred earlier at S1. Based on the satellite data,
it was shown that this part of the lake regularly exhibits higher Chl-a
concentrations in spring (Kiefer et al., 2015).
Given the mesotrophic status of the lake, in March, the euphotic
layer receives an input of nutrients through winter mixing, bringing
up nutrients from the bottom of the lake, which results in reactive phos-
phorus concentrations higher than 10 μg/L in the euphotic zone. In Lake
Geneva, phosphorus is not yet a limiting factor for phytoplankton
growth at this time of the year because most species may be limited
by phosphorus as concentration becomes b10 μg/L (Sas, 1989) and se-
vere reduction in algal growth may occur if soluble reactive phosphorus
concentrations fall below 3 μg/L (Grover, 1989;Suttle et al., 1988).
Moreover, according to the conceptual model of the Plankton Ecology
Group (PEG) (Sommer et al., 2012, 1986)andReynolds (1997) in
such deep lakes, light is the critical resource for phytoplankton growth
in early spring. Observations conrm that phytoplankton developed in
areas of Lake Geneva where our simulations indicated that the thermal
stratication sets in earlier in a more sheltered area compared to the
rest of the lake. Indeed, our results are coherent with the PEG model
which states that thermal stratication triggers the algal development
in spring in deep lakes due to enhanced light availability for phyto-
plankton growth resulting from restrictions on the mixing depth.
This concept has long been described as Sverdrup's critical mixing
depth, i.e. phytoplankton is able to grow only if the mixing depth is
less than a critical depth so that net photoautotrophic growth is possible
(Sverdrup, 1953). However, this concept is not always in line with ob-
servations since phytoplankton blooms have been observed preceding
the onset of stratication, when mixing is still unrestricted (Eilertsen,
1993). Huisman et al., (1999) added the concept of critical turbulence
to the concept of critical mixing depth. In deep and relatively clear
lakes, phytoplankton can maintain development, irrespective of mixing
depth if their growth rate in the upper layer exceeds vertical mixing
rates. The sheltered area in the eastern part of Lake Geneva may favor
early bloom development following either one of these concepts, either
by reducing mixing depth or by reducing turbulent mixing rates to
below the respective critical values.
Upwelling event in summer 2009
Upwelling events were previously observed in some regions of Lake
Geneva (Oesch et al., 2008) and more generally in large lakes (Plattner
Fig. 11. Second period of interest analyzed in spring 2010, a) Times series of wind speed
(U10) and direction (α10), b) Contour plots of simulated water temperature (WT) at the
fourstations,S1,S2,S3andS4,andc)Comparisonofbuoyancyfrequency(N
2
)acrossthelake.
Fig. 12. Third period of interest analyzed in spring 2011, a) Times series of wind speed (U10)
and direction (α10), b) Contour plots of simulated water temperature (WT) at the four
stations, S1, S2, S3 and S4, and c) Comparison of buoyancy frequency (N
2
)acrossthelake.
762 F. Soulignac et al. / Journal of Great Lakes Research 44 (2018) 756764
et al., 2006). When characterizingthe thermal structure of Lake Geneva
using the 3D model, we found that we could link satellite images and
simulation results for the upwelling event of September 4, 2009. In
situ measurements showed that phytoplankton developed at about
15 m depth at that time. We showed that surface Chl-aheterogeneities
observed on September 7, 2009, were related to this upwelling event.
The model suggests a displacement of deep phytoplankton to the sur-
face and provides additional evidence of therole of upwelling on phyto-
plankton abundance heterogeneity in lakes (Cyr, 2017;Huang et al.,
2014). At this stage, our model does not include nutrients or biology,
so it cannot attest for an effective supply of nutrients to the upper wa-
ters that will stimulate algae productivity and phytoplankton growth.
So, it remains unresolved whether the upwelling event indeed en-
hanced primary production and phytoplankton growth as is sometimes
the case during upwelling events (Poschke et al., 2015).
However, this effective enhancement of algal production needs to be
accurately tested because upwelling events are characterized as a tem-
poral displacement of near shore water masses which returned at the
end of each event(Haffner et al., 1984). Therefore, depending on the
time scale, intensity and duration of the event, phytoplankton in the
upper layer should receive little or no benet from the upwelled nutri-
ent rich hypolimnetic waters(Haffner et al., 1984). As a consequence, it
is important to understand and assess when upwellings are efcient fer-
tilizers of the euphotic zone as it has been recorded in various ecosys-
tems (Planas and Paquet, 2016;Valipour et al., 2016). In addition, we
should consider that special attention needs to be paid to the horizontal
mixing occurring during and just after the upwelling event. Coupling
our hydrodynamic model to a biological module should most likely
allow us to estimate the impact of new nutrient availability in the sur-
face waters. Such fertilization of the euphotic zone by nutrient enriched
deep water might have an important role in the functioning of mesotro-
phic lakes. It is important to better assess if these hydrodynamic events
can efciently sustain primary production and how they impact the out-
come of inter-specic competition within the phytoplankton community.
Improving our understanding on the impact of these hydrodynamic
events on phytoplankton communities, might help to understand exem-
plary patterns observed during re-oligotrophication of deep lakes such as
changes in taxonomic composition (Jeppesen et al., 2005)orhysteresisin
phytoplankton biomass (Tadonleke et al., 2009).
Furthermore, several undesirable phytoplankton species are known
to develop at depth in alpine lakes. Mougeotia gracillima or Planktothrix
rubescens develop and proliferate at 1520 m depth near the thermo-
cline in Lake Geneva (Jacquet et al., 2014b;Tapolczai et al., 2015). In
the case of M. gracillima which is not toxic, an upwelling event should
not cause any potential health risks for leisure activities or production
of drinking water. In contrast, if upwelling brings potentially toxic spe-
cies such as P. rubescens to thelake surfacewhere the contact with peo-
ple is more intense than when it remains at greater depth, it may cause
serious health issues (Ibelings et al., 2014). For that reason, in lakes sub-
ject to harmful algal bloom development in deeper layers, it should be
relevant to improve our forecasting of upwelling events (Plattner et
al., 2006) and we suggest the addition of more in situ sampling of phy-
toplankton in these areas for identication of dominant species. 3D
models and remote sensing products could help in identifying the
areas where best to perform the sample collection.
Contribution of 3D models in the interpretation of spatial observations of
Chl-a concentration from Satellite Data
Optical satellite remote sensing has been used to retrieve patterns of
surface Chl-aconcentration in lakes (Bresciani et al., 2011;Kiefer et al.,
2015;Odermatt et al., 2012;Palmer et al., 2015). But this technique is
not capable of detecting deep Chl-aconcentration maxima which are
common in deep lakes undergoing re-oligotrophication (Anneville and
Leboulanger, 2001;Leach et al., 2018). This is a fundamental limitation
of the use of satellite images in characterizing phytoplankton blooms in
lakes. In order to improve Chl-aretrieval, 3D hydrodynamic models
coupled to ecological models could be used to give information about
the Chl-astructure in the vertical dimension which may vary between
different portions of the lake.
Conclusion
A 3D hydrodynamic model of Lake Geneva was created. It is based on
the Delft3D suite software and includes the main tributary (Rhône River)
and 2D high-resolution meteorological forcing. It provides 3D maps of WT
with a 1 h time step on a 1 km horizontal grid size and with a vertical res-
olution of 1 m near the surface to 7 m at the bottom of the lake. This
model was validated using in situ measurements taken at two sampling
stations in the lake over 5 years, from 2008 to 2012. Model performances
were satisfactory, compared to the literature. Here, we have shown that
such a model can be used to detail the lake thermal structure during
two periods, one in spring and another in summer, when surface Chl-a
heterogeneities were observed by satellite data. Our model performed
well in providng explanations for these two horizontal heterogeneities
in Chl-aabundances. For instance, we highlighted the important role of
wind in determining surface phytoplankton abundance. A short and in-
tense wind event in spring is able to create clear surface heterogeneities
in phytoplankton distribution which continue through time. Also in sum-
mer, the role of the wind is very important. It can generate upwelling
events and create surface Chl-aheterogeneities bringing potentially
toxic species to the surface. If we look to the future, the validity of this ap-
proach and the possibility to export this combination of techniques (3D
model and remote sensing) to other study areas in the world could be
considered a fruitful approach, as for example Pinardi et al. (2015) who
effectively used the combination of a hydrodynamic model and of remote
sensing derived products to assess potential algal bloom. Also, a coupled
biological model to the hydrodynamic model could further help to under-
stand better the phytoplankton dynamic and heterogeneities observed in
Lake Geneva.
Acknowledgements
This work was supported the French Agency for Biodiversity (AFB,
before French National Agency for Water and Aquatic Environments
(ONEMA)), ModeL contract N° 15000239, and by the European Space
Agency (ESA) Scientic Exploitation of Operational Missions element
(SEOM), CORESIM contract N° A0/1-8216/15/I-SBo. We wish to thank
the French Alpine Lakes Observatory (SOERE-OLA) and the Interna-
tional Commission for the Protection of Lake Geneva (CIPEL). Data were
from © SOERE OLA-IS, AnaEE-France, INRA Thonon-les-Bains, CIPEL
[2016], developed by Eco-Informatics ORE INRA Team. We also wish
to thank the Department of environment, transport and agriculture
(DETA) of the Geneva water ecology service for providing additional
in situ data as well as Isabel Kiefer for satellite data. Finally, we would
like to thank Tineke Troost and Hans Los (Deltares) for productive dis-
cussions and Pierre Keraudren for English editing.
References
Anneville, O., Leboulanger, C., 2001. Long-terms changes in the vertical distribution of
phytoplankton biomass and primary production in Lake Geneva: a response to the
oligotrophication. Atti Associazione Italiana Oceanologia Limnologia 14, 2535.
Arhonditsis, G.B., Winder, M., Brett, M.T., Schindler, D.E., 2004. Patterns and mechanisms
of phytoplankton variability in Lake Washington (USA). Water Res. 38:40134027.
https://doi.org/10.1016/j.watres.2004.06.030.
Bresciani, M., Stroppiana, D., Odermatt, D., Morabito, G., Giardino, C., 2011. Assessing re-
motely sensed chlorophyll-afor the implementation of the water framework direc-
tive in European perialpine lakes. Sci. Total Environ. 409:30833091. https://doi.
org/10.1016/j.scitotenv.2011.05.001.
Burchard, H., Baumert,H., 1995. On the performance of a mixed-layer modelbased on the
κ-εturbulence clos ure. J. Geophys. Res. Oceans 100:85 238540. https: //doi.org/
10.1029/94jc03229.
Chanudet, V., Fabre, V., van der Kaaij, T., 2012. Application of a three-dimensional hydro-
dynamic model to th e Nam Theun 2 reservoir (Lao PDR). J. Great Lakes Res. 38:
260269. https://doi.org/10.1016/j.jglr.2012.01.008.
763F. Soulignac et al. / Journal of Great Lakes Research 44 (2018) 756764
Chen, Y.W.,Fan, C.X., Teubner, K., Dokulil, M., 2003. Changes of nutrients and phytoplank-
ton chlorophyll-alpha in a large shallow lake, Taihu, China: an 8-year investigation.
Hydrobiologia 506:273279. https://doi.org/10.1023/b:hydr.0000008604.09751.01.
Cyr, H., 2017. Winds and the distribution of nearshore phytoplankton in a stratied lake.
Water Res. 122:114127. https://doi.org/10.1016/j.watres.2017.05.066.
Deltares, 2014. Delft3D-FLOW. User Manual.
Eilertsen, H.C., 1993. Spring blooms and stratication. Nature 363:24. https://doi.org/
10.1038/363024a0.
EuropeanCommission, 2000. Directive 2000/60/EC of the European Parliament andof the
Council of 23 October 2000 establishing a framework for community action in the
eld of water policy. Off. J. Eur. Communities 2000.
Gons, H.J., Auer, M.T., Efer, S.W., 2008. MERIS satellite chlorophyll mapping of oligotro-
phic and eutrophic waters in the Laurentian Great Lakes. Remote Sens . Environ.
112:40984106. https://doi.org/10.1016/j.rse.2007.06.029.
Grover, J.P., 1989. Phosphorus-dependant growth kinetics of 11 species of fres hwater
algae. Limnol. Oceanogr. 34:341348. https://doi.org/10.4319/lo.1989.34.2.0341.
Haffner, G.D., Yallop, M.L., Hebert, P.D.N., Grifths, M., 1984. Ecological signicance of up-
welling events in Lake Ontario. J. Great Lakes Res. 10:2837. https://doi.org/10.1016/
s0380-1330(84)71804-1.
Huang, J.C., Gao, J.F., Hormann, G.,Fohrer, N., 2014. Modeling the effects of environmental
variables on short-term spatial changes in phytoplankton biomass in a large shallow
lake, Lake Taihu. Environ. Earth Sci. 72:36093621. https://doi.org/10.1007/s12665-
014-3272-z.
Huisman, J., van Oostveen, P., Weissing, F.J., 1999. Critical depth and critical turbulence:
two different mechanisms for the development of phytoplankton blooms. Limnol.
Oceanogr. 44:17811787. https://doi.org/10.4319/lo.1999.44.7.1781.
Ibelings, B.W., Backer, L.C., Kardinaal, W.E.A ., Chorus, I., 2014. Current approaches to
cyanotoxin risk assessment and risk management around the globe. Harmful Algae
40:6374. https://doi.org/10.1016/j.hal.2014.10.002.
Ibelings, B.W., Vonk, M., Los, H.F.J., van der Molen, D.T., Mooij, W.M., 2003. Fuzzy model-
ing of cyanobacterial surface waterblooms: validation with NOAA-AVHRR satellite
images. Ecol. Appl. 13:14561472. https://doi.org/10.1890/01-5345.
Jacquet, S., Domaizon, I., Anneville, O., 2014a. The need for ecologicalmonitoring of fresh-
waters in a changing world: a case study of Lakes Annecy, Bourget, and Geneva. En-
viron. Monit. Assess. 186:34553476. https://doi.org/10.1007/s10661-014-3630-z.
Jacquet, S., Kerimoglu, O., Rimet, F., Paolini, G., Anneville, O., 2014b. Cyanobacterial bloom
termination: the disappearance of Planktothrix rubescens from Lake Bourget (France)
after restoration. Freshw. Biol. 59:24722487. https://doi.org/10.1111/fwb.12444.
Jeppesen, E., Sondergaard, M., Jensen, J.P., Havens, K.E., Anneville, O., Carvalho, L.,
Coveney, M.F., Deneke, R., Dokulil, M.T., Foy, B., Gerdeaux, D., Hampton, S.E., Hilt, S.,
Kangur, K., Kohler, J., Lammens, E., Lauridsen, T.L., Manca, M., Miracle, M.R., Moss,
B., Noges, P., Persson, G., Phillips, G., Portielje, R., Schelske, C.L., Straile, D., Tatrai, I.,
Willen, E., Winder, M., 2005. Lake responses to reduced nutrient loading - an analysis
of contemporary long-term data from 35 case studies. Freshw. Biol. 50:17471771.
https://doi.org/10.1111/j.1365-2427.2005.01415.x.
Kacikoc, M., Beyhan, M.,2014. Hydrodynamic and water quality modeling of Lake Egirdir.
Clean: Soil Air Water 42:15731582. https://doi.org/10.1002/clen.201300455.
Kauer, T., Kutser, T., Arst, H., Danckaert, T., Noges, T., 2015. Modelling primary production
in shallow well mixed lakes based on MERIS satellite data. Remote Sens. Environ.
163:253261. https://doi.org/10.1016/j.rse.2015.03.023.
Kiefer, I., Odermatt, D., Anneville, O., Wuest, A., Bouffard, D., 2015. Application of remote
sensing for the optimization of in-situ sampling for monitoring of phytoplankton
abundance in a large lake. Sci. Total Environ. 527:493506. https://doi.org/10.1016/
j.scitotenv.2015.05.011.
Leach, T.H., Beisn er, B.E., Carey, C.C., Pernica, P., Ros e, K.C., Huot, Y., Brentrup, J.A.,
Domaizon, I., Grossart, H.-P., Ibelings, B.W., Jacquet, S., Kelly, P.T., Rusak, J .A.,
Stockwell, J.D., Straile, D., Verburg, P., 2018. Patterns and drivers of deep chlorophyll
maxima structure in 100 lakes: the relative importance of light and thermal strati-
cation. Limnol. Oceanogr. 63:628646. https://doi.org/10.1002/lno.10656.
Lemmin, U., 1995. Limnologie physique. In: Pourriot, R., Meybeck, M. (Eds.), Limnologie
Générale. Masson, Paris, pp. 60114.
Leoni, B., Marti, C.L., Imberger, J., Garibaldi, L., 2014. Summer spatial variations in phyto-
plankton composition and biomass in surface waters of a warm-temperate, deep,
oligoholomictic lake: Lake Ise o, Italy. Inland Waters 4:303310. https://doi.org/
10.5268/iw-4.3.569.
Li, Z.J., Chen, Q.W., Xu, Q., 2015. Modeling algae dynamics in Meiliang Bay of Taihu Lake
and parameter sensitivity analysis. J. Hydro Environ. Res. 9:216225. https://doi.
org/10.1016/j.jher.2014.10.001.
Mccombs, M.P., Mulligan, R.P., Boegman, L., 2014. Offshore wind farm impacts on surface
waves and circulation in Eastern Lake Ontario. Coast. Eng. 93:3239. https://doi.org/
10.1016/j.coastaleng.2014.08.001.
Moreno-Ostos, E., da Silva, S.L.R., de Vicente, I., Cruz-Pizarro, L., 2007. Interannual and be-
tween-site variability in the occurrence of clear water phases in two shallow Mediter-
ranean lakes. Aquat. Ecol. 41:285297. https://doi.org/10.1007/s10452-006-9072-0.
Odermatt, D., Pomati, F., Pitarch, J., Carpenter, J., Kawka, M., Schaepman, M., Wuest, A.,
2012. MERIS observations of phytoplankton blooms in a stratied eutrophic lake. Re-
mote Sens. Environ. 126:232239. https://doi.org/10.1016/j.rse.2012.08.031.
Oesch, D., Jaquet, J.M., Klaus, R., Schenker, P., 2008. Multi-scale thermal pattern monitor-
ing of a large lake (Lake Geneva) using a multi-sensor approach. Int. J. Remote Sens.
29:57855808. https://doi.org/10.1080/01431160802132786.
Palmer, S.C.J., Odermatt, D., Hunter, P.D., Brockmann, C., Presing, M., Balzter, H., Toth, V.R.,
2015. Satellite remote sensing of phytoplankton phenology in Lake Balaton using 10
years of MERIS observations. Remote Sens. Environ. 158:441452. https://doi.org/
10.1016/j.rse.2014.11.021.
Pelechaty, M., Owsianny, P.M., 2003. Horizontal distribution of phytoplankton as related to
the spatial heterogeneity of a lake - a case study from two lakes of the Wielkopolski
National Park (western Poland). Hydrobiologia 510:195205. https://doi.org/10.1023/
B:HYDR.0000008644.67963.c4.
Pinardi, M., Fenocchi, A., Giardino, C., Sibilla, S., Bartoli, M., Bresciani, M., 2015. Assessing po-
tential algal blooms in a shallow uvial lake by combining hydrodynamic modelling and
remote-sensed images. Water 7 (5):19211942. https://doi.org/10.3390/w7051921.
Planas, D.,Paquet, S., 2016. Importance of climate change-physicalforcing on the increase
of cyanobacterial blooms in a small, stratied lake. J. Limnol. 75:201214. https://doi.
org/10.4081/jlimnol.2016.1371.
Plattner, S., Mason, D.M., Leshkevich, G.A., Schwab, D.J., Rutherford, E.S., 2006. Classifying
and forecasting coastal upwellings in Lake Michigan using satellite derived tempera-
ture images and buoy data. J. Great Lakes Res. 32:6376. https://doi.org/10.3394/
0380-1330(2006)32[63:cafcui]2.0.co;2.
Pomati, F.,Jokela, J., Simona, M., Veronesi, M., Ibelings, B.W., 2011.An automated platform
for phytoplankton ecology and aquatic ecosystem monitoring. Environ. Sci. Technol.
45:96589665. https://doi.org/10.1021/es201934n.
Poschke, F., Lewandowski, J., Engelhardt, C., Preuss, K., Oczipka, M., Ruhtz, T., Kirillin, G.,
2015. Upwelling of deep water during thermal stratication onset-a major mecha-
nism of vertical transport in small temperate lakes in spring? Wat er Resour. Res.
51:96129627. https://doi.org/10.1002/2015wr017579.
Razmi, A.M., Barry, D.A., Bakhtyar, R., Le Dantec, N., Dastgheib, A., Lemmin, U., Wuest, A.,
2013. Current variability in a wide and open lacustrine embayment in Lake Geneva
(Switzerland). J. Great Lakes Res. 39:455465. https://doi.org/1 0.1016/j.
jglr.2013.06.011.
Razmi, A.M., Barry, D.A., Lemmin, U., Bonvin, F., Kohn, T., Bakhtyar, R., 2014. Direct effects
of dominant winds on residence and travel times in the wide and open lacustrine
embayment: Vidy Bay (Lake Geneva, Switzerland). Aquat. Sci. 76:S59S71. https://
doi.org/10.1007/s00027-013-0321-8.
Razmi, A.M., Lemmin, U., Bouffard, D., Wuest, A., Uittenbogaard, R.E., Barry, D.A., 2017.
Gyre formation in open and deep lacustrine embayments: the example of Lake Ge-
neva,Switzerland.Environ.FluidMech.17:415428. https://doi.org/10.1007/
s10652-016-9494-8.
Read, J.S., Hamilton,D.P., Jones, I.D., Muraoka, K., Winslow, L.A.,Kroiss, R., Wu, C.H.,Gaiser,
E., 2011. Derivation of lakemixing and stratication indicesfrom high-resolution lake
buoy data. Enviro n. Model. Softw. 26:13251336. htt ps://doi.org/10.1016/j.
envsoft.2011.05.006.
Reynolds, C.S., 1997. Vegetation process in the pelagic: a model for ecosystem theory. In:
Kinne, O. (Ed.), Excellence in Ecology. ECI, Oldendorf.
Sas, H., 1989. Lake Restoration by Reduction of Nutrient Loading: Expectations, Experi-
ences, Extrapolations. Acdemia Verlag Richarz, St. Augustin.
Schroeder, T., Behnert, I., Schaale, M., Fischer, J., Doerffer, R., 2007.Atmospheric correction
algorithm for MERIS abovecase-2 waters. Int. J. Remote Sens. 28:14691486. https://
doi.org/10.1080/01431160600962574.
Sommer,U.,Adrian,R.,Domis,L.D.,Elser,J.J.,Gaedke,U.,Ibelings,B.,Jeppesen,E.,Lurling,M.,
Molinero, J.C., Mooij, W.M., van Donk, E., Winder, M., 2012. Beyond the plankton ecol-
ogy group (PEG) model: mechanisms driving plankton succession. Annu. Rev. Ecol.
Evol. Syst. 43 (43):429448. https://doi.org/10.1146/annurev-ecolsys-110411-160251.
Sommer, U., Gliwicz, Z.M., Lampert,W., Duncan, A., 1986. ThePEG-model of seasonal suc-
cession of planktonic events in fresh waters. Arch. Hydrobiol. 106, 433471.
Soulignac, F., Vincon-Leite, B., Lemaire, B.J., Martins, J.R.S., Bonhomme, C., Dubois, P.,
Mezemate, Y., Tchiguirinskaia, I., Schertzer, D., Tassin, B., 2017. Performance assess-
ment of a 3D hydrodynamic model using high temporal resolution measurements
in a shallow urban La ke. Environ. Model. Assess. 22:309322. https://doi.org/
10.1007/s10666-017-9548-4.
Strickland, J.D.H., Parsons, T.R., 1968. A pratical handbook for seawater analysis. Fisheries
research board, Ottawa, Canada.
Stewart, T.J., Rudstam, L., Watkins, J., Johnson, T.B., Weidel, B., Koops, M.A., 2016.Research
needs to better understand Lake Ontario ecosystem function: a workshop summary.
J. Great Lakes Res. 42:15. https://doi.org/10.1016/j.jglr.2015.10.017.
Suttle, C.A., Stockner, J.G., Shortreed, K.S., Harrison, P.J., 1988. Time-courses of size-frac-
tionated phosphote uptake: are larger cells better competitors for pulse of phosphate
than smaller cells. Oecologia 74:571576. https://doi.org/10.1007/bf00380055.
Sverdrup,H.U., 1953. On conditions for the vernal bloomingof phytoplankton.ICES J. Mar.
Sci. 18:287295. https://doi.org/10.1093/icesjms/18.3.287.
Tadonleke, R.D., Lazzarotto, J., Anneville, O., Druart, J.C., 2009. Phytoplankton productivity
increased in Lake Geneva despite phosphorus loading reduction. J. Plankton Res. 31:
11791194. https://doi.org/10.1093/plankt/fbp063.
Tapolczai, K., Anneville, O., Padisak, J., Salmaso, N., Morabito, G., Zohary, T., Tadonleke, R.
D., Rimet, F., 2015. Occurrence and mass development of Mougeotia spp.
(Zygnemataceae) in large, deep lakes. Hydrobiologia 745:1729. https://doi.org/
10.1007/s10750-014-2086-z.
Valipour, R., Leon, L.F., Depew, D., Dove, A., Rao, Y.R., 2016. High-resolution modeling for
development of nearshore ecosystem objectives in eastern Lake Erie. J. Great Lakes
Res. 42:12411251. https://doi.org/10.1016/j.jglr.2016.08.011.
Viljanen, M., Holopainen, A.L., Rahkola-Sorsa, M., Avinsky, V., Ruuska, M., Leppanen, S.,
Rasmus, K., Vouti lainen, A., 2009. Temporal and spatial heterogeneity o f pelagic
plankton in Lake Pyhaselka, Finland. Boreal Environ. Res. 14, 903913.
Wahl, B., Peeters, F., 2014. Effect of climatic changes on stratication and deep-water r e-
newal in Lake Constance assessed by sensitivity studies with a 3D hydrodynamic
model. Limnol. Oceanogr. 59:10351052. https://doi.org/10.4319/lo.2014.59.3.1035.
Xu, F.L., Tao, S., Dawson,R.W., Li, P.G., Cao, J., 2001. Lake ecosystem health assessment: in-
dicators and methods. Water Res. 35:31573167. https://doi .org/10.1016/s0043-
1354(01)00040-9.
764 F. Soulignac et al. / Journal of Great Lakes Research 44 (2018) 756764
... Recently, several studies have been conducted to better understand the spatio-temporal heterogeneities of phytoplankton abundance in large lakes. Dynamics of algal growth can be explained by temporal and spatial variability in thermal stratification dynamics and internal wave motions (Yang et al. 2016;Soulignac et al. 2018). Furthermore, a higher abundance of phytoplankton has been reported around river inflow areas (Larson et al. 2013;Kiefer et al. 2015;Soomets et al. 2019). ...
... Using satellite images and a hydrodynamic model, Soulignac et al. (2018) indicated an earlier onset of phytoplankton development in spring 2011 in the Haut Lac compared to the rest of the lake. They argued that this sheltered area may favor an earlier stratification and hence improved access to light, leading to an earlier onset of the spring bloom. ...
Article
Full-text available
River–lake transition zones have been identified as major drivers of phytoplankton growth. With climate change reducing the frequency of complete lake overturns, it is expected that the Rhône River, the main tributary to Lake Geneva (France/Switzerland), will become the major source of nutrients for the lake euphotic zone. The river–lake transition zone was hence examined at the mouth of the Rhône River with the aim of understanding the complexities and controls of phytoplankton distribution in this specific deltaic ecosystem. Two field campaigns were carried out in which water samples were collected from longitudinal and transversal transects across the transition zone. These samples were analyzed for both nutrient and phytoplankton concentrations, while the fraction of Rhône River water in a lake sample was determined by the stable isotope composition of the water. The results indicate contributions in P and Si related to the Rhône intrusion into the lake. Furthermore, this river–lake transition zone appears to be a dynamic area that can locally present optimal conditions for phytoplankton growth. In early spring, a wind event broke the early and weak stratification of the lake, forcing the Rhône River and its turbidity plume to intrude deeper. Thus, this sharp drop of the turbidity within the euphotic zone allowed an increase in the phytoplankton biovolume of 44%. In early fall, outside of the turbid near field of the river mouth, the Rhône interflow, located just below the thermocline, promoted a local deep chlorophyll maximum.
... sensing also allows researchers and scientists to rapidly and accurately obtain vast amounts of satellite data at high resolution. This tool has become very useful for fluvial hydrology and geomorphology to delineate and characterise river channels or to classify and monitor the changes in river landscape (Soulignac et al., 2018). Note that, unlike traditional remote sensing with multiple spectral sensors that have a limited number of broad spectral bands and were designed mainly to detect the concentration of the primary pigment in phytoplankton, hyperspectral remote sensing allows researchers to collect images across the full spectrum of visible and infrared light. ...
... From this point of view, the transparency and simplicity of models, such as decision trees and fuzzy models with simple if-then rules, are valuable assets. These models have proven their success in many fields, such as medicine (Ahmadi et al., 2018), electronics (Singh et al., 2013), (waste)water treatment (Porro et al., 2018) and climate change (Ho et al., 2021). On the other hand, having limited practical experience, researchers advance new statistical techniques that have appeared difficult for practitioners to learn due to their lack of participation in the development of these techniques (Corominas et al., 2017). ...
Article
Full-text available
1. As one of the earth's key ecosystems, rivers have been intensively studied and modelled through the application of machine learning (ML). With the amount of large data available, these computer algorithms are ever increasing in numerous fields, although there is ongoing scepticism and scholars still question the actual impact and deliverables of algorithms. 2. This study aims to provide a systematic review of the state-of-the-art ML-based techniques, trends, opportunities and challenges in river research by applying text mining and automated content analysis. 3. Unsupervised and supervised learning have dominated river research while neural networks and deep learning have also gradually gained popularity. Matrix factorisation and linear models have been the most popular ML algorithms, with around 1300 and 800 publications on these topics in 2020 respectively. In contrast, river researchers have had few applications in multiclass and multilabel algorithm, associate rule and Naïve Bayes. 4. The current article proposes an end-to-end workflow of ML applications in river research in order to tackle major ML challenges, including four steps: (1) data collection and preparation; (2) model evaluation and selection; (3) model application; and (4) feedback loops. Within this workflow, river modellers have to balance numerous trade-offs related to model traits, such as complexity, accuracy, interpretability, bias, data privacy and accessibility and spatial and temporal scales. Any choices made when balancing the trade-offs can lead to different model outcomes affecting the final applications. Hence, it is necessary to carefully consider and specify modelling goals, understand the data collected and maintain feedback loops in order to continuously improve model performance and eventually reach the research objectives. Moreover, it remains crucial to address the users' needs and demands that often entail additional elements, such as computational cost, development time and the quantity, quality and compatibility of data. Furthermore, river researchers should account for new technologies and regulations in data collection and protection that are transforming the development and applications of ML, most notably data warehouse and information management with multiple-cycles that are becoming a cornerstone of the integration of ML in decision-making in river and ecosystem management.
... Figure 3.13 illustrates OLCI BNN chla through prior C2RCC AC and associated uncertainty products over Lake Geneva from March 2020, November 2020, and August 2021. The maps using the BNN match the time series chla patterns and ranges for Lake Geneva from the oligotrophic states in March 2020 to mesotrophic levels in August 2021 (Soulignac et al., 2018). ...
Thesis
Full-text available
Phytoplankton constitute the bottom of the aquatic food web, produce half of Earth’s oxygen and are part of the global carbon cycle. A measure of aquatic phytoplankton biomass therefore functions as a biological indicator of water status and quality. The abundance of phytoplankton in most lakes on Earth is low because they are weakly nourished (i.e., oligotrophic). It is practically infeasible to measure the millions of oligotrophic lakes on Earth through field sampling. Fortunately, phytoplankton universally contain the optically active pigment chlorophyll-a, which can be detected by optical sensors. Earth-orbiting satellite missions carry optical sensors that provide unparalleled high spatial coverage and temporal revisit frequency of lakes. However, when compared to waters with high nutrient loading (i.e., eutrophic), the remote sensing estimation of phytoplankton biomass in oligotrophic lakes is prone to high estimation uncertainties. Accurate retrieval of phytoplankton biomass is severely constrained by imperfect atmospheric correction, complicated inherent optical property (IOP) compositions, and limited model applicability. In order to address and reduce the current estimation uncertainties in phytoplankton remote sensing of low - moderate biomass lakes, machine learning is used in this thesis.
... Fig. 13 illustrates OLCI BNN chla through prior C2RCC AC and associated uncertainty products over Lake Geneva from March 2020, November 2020, and August 2021. The maps using the BNN match the time series chla patterns and ranges for Lake Geneva from the oligotrophic states in March 2020 to mesotrophic levels in August 2021 (Soulignac et al., 2018). The retrieval uncertainties ranged between 20 and 60% in Lake Geneva similar to the results of the previous match-up and time series assessments. ...
Article
Full-text available
Satellite remote sensing of chlorophyll-a concentration (chla) in oligotrophic and mesotrophic lakes faces uncertainties from sources such as atmospheric correction, complex inherent optical property compositions, and imperfect algorithmic retrieval. To improve chla estimation in oligo- and mesotrophic lakes, we developed Bayesian probabilistic neural networks (BNNs) for the Sentinel-3 Ocean and Land Cover Instrument (OLCI) and Sentinel-2 MultiSpectral Imager (MSI). The BNNs were built using an in situ dataset of oligo- and mesotrophic water bodies (1755 observations from 178 systems; median chla: 5.11 mg m⁻³, standard deviation: 10.76 mg m⁻³) and provide a per-pixel uncertainty percentage associated with retrieved chla. Shifts of oligo- and mesotrophic systems into the eutrophic regime, characterised by higher biomass levels, are widespread. To account for phytoplankton biomass fluctuation, a set of eutrophic lakes (167 observations from 31 systems) were included in this study (maximum chla 68 mg m⁻³). The BNNs were evaluated through five assessments including single day and time series match-ups with OLCI and MSI. OLCI BNN accuracy gains of >25% and MSI BNN accuracy gains of >15% were achieved in the assessments when compared to chla reference algorithms for oligotrophic waters (chla ≤ 8 mg m⁻³). In comparison to the reference algorithms, the accuracy gains of the BNNs decreased as chla and trophic levels increased. To measure the quality of the provided BNN uncertainty estimate, we calculated the prediction interval coverage probability (PICP), Sharpness and mean absolute calibration difference (MACD) metrics. The associated BNN chla uncertainty estimate included the reference in situ chla values for most observations (PICP ≥ 75%) across the different performance assessments. Further analysis showed that the BNN chla uncertainty estimate was not constantly well-calibrated across different evaluation strategies (Sharpness 1.7–6, MACD 0.04–0.25). BNN uncertainties were used to test two chla improvement strategies: 1) identifying and filtering uncertain chla estimates using scene-specific thresholds, and 2) selecting the most accurate prior atmospheric correction algorithm per individual satellite observation to retain chla with the lowest BNN uncertainty. Both strategies increased the quality of the chla result and demonstrated the significance of uncertainty estimation. This study serves as research on Bayesian machine learning for the estimation and visualisation of chla and associated retrieval uncertainty to develop harmonised products across OLCI and MSI for small and large oligo- and mesotrophic lakes.
... More classical approaches include mathematical modelling of different complexity from simple linear to multilevel models. A number of 3D models has been elaborated for studies of seasonal and spatial dynamics of the key physical, chemical and biological processes in large lakes worldwide (Leon et al., 2012;Bocaniov et al., 2016;Soulignac et al., 2018;Zou et al., 2020). Lake Peipsi is not an exception. ...
Article
Full-text available
Lake Peipsi, one of the world's largest lakes, is shared between Estonia and Russia. The water quality in different parts of the lake has so far been assessed independently. Here we explore opportunities for combining data of Estonian and Russian monitoring. For that, we 1) analysed the compatibility of data for some water quality variables; 2) estimated the potential effects of the differences in sampling frequency; 3) provided a few regression models to calculate the missing data for months not sampled by the Russian side. Data of the concurrent Estonian and Russian sampling indicated a good compatibility. Estonian data analysis suggested that water quality assessment results are sensitive to sampling frequency. For example , total phosphorus (TP) in the largest basin showed a long-term decreasing trend in three month data that disappeared when data for other months were added. Disregarding some months may lead to under-or overestimation of certain factors with no consistency in the response of different basins. Hence, data of the whole ice-free period are recommended for an adequate water quality assessment. Furthermore, we demonstrated that monthly values of the water quality variables of the same year are autocorrelated. Based on this, we filled the gaps in the long-term data and compiled a dataset for the whole lake that enables its most comprehensive use in water quality assessment and management. Long-term data revealed no water quality improvement of Lake Peipsi. Further reduction of the external nutrient load is needed. Eutrophication is sustained by high internal phosphorus load.
... It comprises, among others, of two Sentinel-2 satellites with 10 to 60 m spatial resolution and 5-day joint revisit time (Drusch et al., 2012) and two Sentinel-3 satellites with 300 m spatial resolution and daily joint revisit (Donlon et al., 2012) for land and ocean monitoring, respectively. While remote sensing is particularly powerful in monitoring the surface layer of lakes (Lehmann et al., 2021), numerical modelling provides a convenient tool to evaluate the dynamics of the entire lake and to rationalize any remotely observed spatial variability Curtarelli et al., 2014;Soulignac et al., 2018;Wynne et al., 2013). It is thus sensible to couple remote sensing with numerical simulation to merge the advantages of each individual method. ...
Article
Propelled by the rapid development of equipment, technology and computational power, the monitoring and simulation of the hydrodynamics in lakes have steadily advanced. In contrast, water quality simulations are more difficult to implement, due to the difficulty in obtaining large-scale, spatially resolved field observations for model validation and the number of interacting processes to be parameterized. Here we demonstrate that remote sensing data can be used to inform Lagrangian particle tracking in a large lake, and vice versa. We used total suspended matter (TSM) as a parameter that can be both estimated from the backscattering in satellite images and modelled in terms of particle abundance. Specifically, we compared TSM concentrations in Lake Geneva deduced from images taken by Sentinel-2 and Sentinel-3 satellites to those estimated from Delft3D hydrodynamic and particle tracking models. TSM concentrations obtained from both methods were compared over a time span of up to 5 days in several scenario studies, including instantaneous and continuous point sources and large-scale TSM simulations. The results demonstrate that remote sensing images can serve to calibrate and validate particle tracking models with independent observations. The model was able to capture both the position of a TSM cloud arising 5 days after an instantaneous point source release, and the direction of particle transport and TSM plume size resulting from a continuous source. Even when simulating the whole lake domain, model results closely approximated the satellite-derived TSM concentrations along lake transects within 9%. In return, the particle tracking model was able to complete partially impaired satellite images, and fill in a four-day image gaps between satellite revisits. The synergy of remote sensing techniques and particle tracking modelling allows a rapid, continuous and more accurate analysis on solute transport in lakes.
... Meteorological reanalyses usually cover multi-decadal periods and have the great benefit of being spatialized over vast portions of the globe. Even though their use in limnological studies is quite recent, they have already been used to simulate water temperature (Layden et al. 2016, Piccolroaz et al. 2020), stratification dynamics (Frassl et al. 2018) and phytoplankton distribution (Soulignac et al. 2018). As shown in this work, their use as external forcing to thermal-hydrodynamic models can yield, provided that observations are available for calibration and validation, to accurate simulations of the behaviour of water bodies even in the absence of local meteorological observations. ...
Thesis
The ecological state of freshwater ecosystems worldwide has deteriorated along the past decades. Anthropogenic pressures have altered their physical and biogeochemical dynamics, acting both within their watershed and on the climatic conditions. Eutrophication and climate change contributed to the increase of algal blooms, and in particular of toxic cyanobacteria blooms , which currently constitute one of the main concern in the management of water resources.With the advance of urbanization, an increasing number of lakes are located in metropolitan areas. The high loads of nutrients and pollutants coming from the watershed often lead urban lakes to eutrophic conditions and cyanobacteria blooms, that cause bathing bans and restrictions for aquatic sports. Responsive surveys and long-term climate change impact studies are essential for the management of such sites, but rarely addressed.In this respect, modelling tools are of central importance to better understand the functioning of aquatic ecosystems, the factors triggering harmful algal blooms, and to support the management of water resources. However, aquatic ecological models are often complex and highly parametrized, and their implementation and calibration are challenging. Automated strategies for parameters calibration are available but are rarely applied. Furthermore, data from traditional periodical limnological survey do not allow to test the models on dynamics quicker than the span between two successive campaigns, and to thoroughly assess the uncertainty of their outcomes.In this context, this PhD thesis focuses on the use of deterministic models to reproduce the thermal dynamics and phytoplankton dynamics, notably cyanobacteria, in a small and shallow urban lake on different time-scales. To do so, two coupled hydrodynamic and biogeochemical three-dimensional (3D) models are implemented and analysed. The models used here are the FLOW and BLOOM modules from the Delft3D modelling suite, and the biogeochemical library Aquatic EcoDynamics coupled with the hydrodynamic model TELEMAC3D. The models are applied on Lake Champs-sur-Marne, an urban lake located in the East of Paris that suffers from strong cyanobacterial blooms and for which an extensive data set is available.This work aims to address in detail three strategic elements in lake ecosystem modelling:•The impact of climate change on the thermal regime of small and shallow lakes, and its relation to cyanobacterial growth. This is assessed through long-term 3D hydrodynamic simulations that allowed to hindcast the evolution of the study site during the past six decades.• The applicability and the benefits of automated calibration for complex biogeochemical models. This is done through an innovative methodology for parameter estimation: Approximate Bayesian Computation (ABC), tested here for the first time on a complex, highly-parametrized model.•The coupling and the feedbacks between hydrodynamic and biogeochemical models focusing on different time scales, and the importance of an extensive data set, that includes continuous high-frequency observations.The results show that the impact of climate change on small and shallow lakes can be severe, with consequences on the stratification dynamics and that thermal conditions increasingly favourable for cyanobacterial growth have established over time in the study site. This suggests that cyanobacteria dominance could become a widespread issue in the near future, if such trends are confirmed. Furthermore, this work proves that automated calibration strategies, and ABC in particular, can be profitably applied to complex physically-based biogeochemical models in order to improve their results over the period chosen for calibration. Eventually, this work also highlights the importance of an extensive data set to set-up a coupled 3D hydrodynamic / biogeochemical model, and analyse and exploit its results over different time scales.
Article
Full-text available
We present a Bayesian inference for a three-dimensional hydrodynamic model of Lake Geneva with stochastic weather forcing and high-frequency observational datasets. This is achieved by coupling a Bayesian inference package, SPUX, with a hydrodynamics package, MITgcm, into a single framework, SPUX-MITgcm. To mitigate uncertainty in the atmospheric forcing, we use a smoothed particle Markov chain Monte Carlo method, where the intermediate model state posteriors are resampled in accordance with their respective observational likelihoods. To improve the uncertainty quantification in the particle filter, we develop a bi-directional long short-term memory (BiLSTM) neural network to estimate lake skin temperature from a history of hydrodynamic bulk temperature predictions and atmospheric data. This study analyzes the benefit and costs of such a state-of-the-art computationally expensive calibration and assimilation method for lakes.
Article
The author considers the need for evaluation of territories in order to ensure the ecological balance and stability of the environment. It is possible only when all the features of the terrain are known in detail. One of the ways that contribute to high-quality and detailed analysis is three-dimensional mapping of the land using materials of various survey types, which enables carrying out the necessary actions on analytics, preliminary planning and design under office conditions. The object of the study is a rural area, the choice of which is due to the possibility of providing the general favorable environment in the region, as well as lack of attention to such territories and their “desolation”. The features of various types of software are analyzed; the way of their integration is described. A template for a database of objects of a three-dimensional map was developed. For mapping, the software designed for creating computer games is used, with its qualitative advantages of visualization and a considerable library of objects. The interface, tools, various features and capabilities of the used graphic editor are presented. An original mapping technique was developed; the stages of creating a 3D map are described and illustrated in detail. Further prospects of the work are outlined; its novelty and practical significance are noted.
Article
In the modern world, the ecological situation remains very tense. In this regard, solving the envi-ronmental problems of individual territories is an urgent task. The work is devoted to the application of the cartographic method for studying the ecological situation in the territory of the municipal district of the Altai Region. One of the most effective methods of research is the cartographic representation of the terrain in three-dimensional form, providing ease of perception, facilitating the processes of analysis, planning and design. Some environmental problems in the territory are highlighted, the state of protective forest strips is studied in detail. A 3D map of the actual state of the problem situation has been compiled, solutions have been proposed, and the planned results of project activities are visually displayed. The features of using the map are described; the possibility and necessity of interaction of various spheres and branches of activity in matters of environmental safety are noted.
Article
Full-text available
The vertical distribution of chlorophyll in stratified lakes and reservoirs frequently exhibits a maximum peak deep in the water column, referred to as the deep chlorophyll maximum (DCM). DCMs are ecologically important hot spots of primary production and nutrient cycling, and their location can determine vertical habitat gradients for primary consumers. Consequently, the drivers of DCM structure regulate many characteristics of aquatic food webs and biogeochemistry. Previous studies have identified light and thermal stratification as important drivers of summer DCM depth, but their relative importance across a broad range of lakes is not well resolved. We analyzed profiles of chlorophyll fluorescence, temperature, and light during summer stratification from 100 lakes in the Global Lake Ecological Observatory Network (GLEON) and quantified two characteristics of DCM structure: depth and thickness. While DCMs do form in oligotrophic lakes, we found that they can also form in eutrophic to dystrophic lakes. Using a random forest algorithm, we assessed the relative importance of variables associated with light attenuation vs. thermal stratification for predicting DCM structure in lakes that spanned broad gradients of morphometry and transparency. Our analyses revealed that light attenuation was a more important predictor of DCM depth than thermal stratification and that DCMs deepen with increasing lake clarity. DCM thickness was best predicted by lake size with larger lakes having thicker DCMs. Additionally, our analysis demonstrates that the relative importance of light and thermal stratification on DCM structure is not uniform across a diversity of lake types.
Article
Full-text available
Urban lakes provide many ecosystem services, e.g., flood control, nature protection, coolness island, recreation. Hydrodynamic models will increasingly be used to enhance these benefits. We present the first validation of a three-dimensional (3D) hydrodynamic model on a small shallow lake with high resolution and high frequency measurements. Lake Créteil, France (area 0.4 km2, mean depth 4.5 m, and catchment area 1 km2) is a former gravel pit and now part of a regional park. The model Delft3D-FLOW was calibrated on a one-month period, with continuous measurements of temperature at five depths at the center of the lake and at three depths at two other stations, and with current speed profiles at the centre of the lake. The model was then verified on 18 1-month periods with similar temperature measurements. The model reproduced very well the temperature dynamics, including the alternation between mixing and stratification periods and internal wave patterns. The mean absolute errors over the five depths at the central point remained below 0.55∘C in spring and summer, the most favorable seasons for phytoplankton growth. Horizontal temperature differences, which rose up to 3∘C at the beginning of stratification periods, were also well reproduced, as well as current speeds. These results are very promising for assessing nutrient and pollutant diffusion, settling and resuspension, as well as for understanding how phytoplankton blooms start in small shallow lakes.
Article
Full-text available
Numerical simulations were carried out to investigate gyres within open lacustrine embayments subjected to parallel-to-shore currents. In such embayments, gyre formation occurs due to flow separation at the embayment’s upstream edge. High momentum fluid from the mixing layer between the embayment and offshore flows into the embayment and produces recirculating flow. Systematic numerical experiments using different synthetic embayment configurations were used to examine the impact of embayment geometry. Geometries included embayments with different aspect ratios, depths and embayment corner angles. The magnitudes of the recirculation and turbulent kinetic energy (TKE) in the embayment vary significantly for angles in the range 40°–55°. Embayments with corner angles less than 50° have much stronger recirculation and TKE, other parameters remaining the same. The numerical findings are consistent with gyre formation observed in two embayments located in Lake Geneva, Switzerland, and thus help explain flow patterns recorded in lacustrine shoreline regions.
Article
Full-text available
We develop a high-resolution (600 m) three-dimensional water quality model for Lake Erie capable of resolving predominant physical processes to study nutrient dynamics, with a particular emphasis on the northern nearshore region of Lake Erie's eastern basin. The lake model output in conjunction with the Cladophora growth model (CGM) is used to predict Cladophora growth. The models were validated using extensive nearshore water quality, Cladophora biomass and tissue phosphorus measurements collected during April–September of 2013. Together, the models were used to evaluate the response of nearshore phosphorus concentrations and Cladophora growth due to changes in external phosphorus loading. The water quality model described here was able to resolve the nearshore dominant physical processes that occur within the Cladophora habitat zone (0–8 m depth) and revealed that phosphorus concentrations in the nearshore are governed by a combination of local inputs from the Grand River and exchange with the hypolimnion. Frequent upwelling events driven by favorable winds of 5–10 days period dominated nearshore–offshore exchanges, and in some years (e.g., 2013) can potentially add significant phosphorus (as soluble reactive phosphorus; SRP) to the nearshore during the spring; however, in other years (e.g., 2008), their effects are less important. The model demonstrates that the offshore supply of phosphorus via upwelling is of ecological significance with respect to Cladophora and that both lake-wide and local action may be required to address nuisance blooms of Cladophora in Lake Erie.
Article
Full-text available
The community structure of planktonic cyanobacteria was studied in a dimictic lake in which recurrent summer surface algal blooms have frequently occurred since the beginning of this millennium. In eutrophic-hypereutrophic lakes, epilimnetic cyanobacterial blooms are promoted by increased ambient temperatures and water column thermal stability, which favour the vertical migration of buoyancy-regulating cyanobacteria. Here we propose that intensified external energy (wind) that alters thermocline stability could explain the occurence of heavy blooms in the surface of lakes with low external nutrient loading. Specifically, we hypothesized that: i) in small stratified lakes with low external nutrient sources, cyanobacterial growth primarily occurs near the lake bottom, where phosphorus is more abundant and light is available; ii) we additionally hypothesized that turbulence induced by strong winds increases the amplitude and energy of metalimnetic internal waves and entrains meta- and hypolimnetic water, rich in nutrients and cyanobacteria, into the epilimnion. The study was done in a small lake (45 Ha, maximum and mean depth 7.2 m and 4.3 m, respectively) with mean epilimnetic dissolved phosphorus concentrations ≈ 4 μg L-1 and chlorophyll α ≈ 8 μg L-1. Vertical temperature profiles during the open season were continuously registered using thermistors. Weekly vertical profiles of light transmission, phytoplankton distribution and water chemistry were also taken. On one occasion, these variables were measured throughout a continuous 24 h cycle. Results demonstrated that summer cyanobacterial blooms were dominated by Plankthotrix spp., which began their cycle in late spring at the bottom of the lake, and grew to form dense metalimnetic biomass peaks. Time series analysis of isotherms and the Lake number indicated that internal metalimnetic waves (seiches) were present through the summer. During the diel sampling cycle, we found that medium to strong westerly wind gust events (~5 to >12 m s-1) induced large amplitude internal waves (mainly V2H1 mode) that vertically displaced the isotherms by more than 3.5 m. During this event the top of the metalimnetic algal peak was entrained through the epilimnion, bringing metalimnetic Plankthotrix spp. to the lake surface, modified the deep metalimnion and hypolimnion concentrations of dissolved oxygen, and caused an upsurge in phosphorus. We conclude that algal and nutrient upwelling linked to intermittent deep mixing events, play an important role in supporting summer cyanobacterial blooms in lake Bromont.
Article
Full-text available
The applicability of satellite imagery products from different sensors (AVHRRderived multi-channel sea surface temperature (MCSST), MODIS sea surface temperature (SST) products 5-Min L2 Swath 1 km and Landsat TM band 6 thermal signature) for the comprehensive monitoring of temperature and its temporal patterns over a large lake is tested in this study. The coverage of cloudfree satellite data for Lake Geneva is reported throughout a year and, more specifically, during a 13 day period in summer 2003. In a second step, we demonstrate the feasibility of the AVHRR/MODIS imagery to discern day and night temperature patterns, by generating day and night climatologies and various spatial statistics over the 13 day period. The different day and night surface thermal patterns observed by satellite imagery could be linked to the thermal structure existing in deeper water using the concept of the diurnal decoupled layer. The forcing of the persistent patterns, two warm cores divided by a saddle-shaped cold anomaly, is explained by wind periodicity and insolation conditions. The patterns can be matched to features postulated by findings of different limnologists in the past. Other surface temperature related phenomena such as water upwelling and downwelling and the occurrence of plumes are related to meteorological and hydrological events. The lakewide average lake surface water temperature (LSWT) trends for day and night during the study period are roughly parallel. A sudden loss of stored heat can be explained by episodes of long fetch, synoptic wind (bise) that interrupted the predominant breeze regime.
Article
The distribution of phytoplankton in lakes is notoriously patchy and dynamic, but wind-driven currents and algal buoyancy/motility are thought to determine where algae accumulate. In this study, nearshore phytoplankton were sampled from different parts of a lake basin twice a day for 4–5 consecutive days, in the spring and in late summer, to test whether short-term changes in phytoplankton biomass and community composition can be predicted from wind-driven currents. On windy days, phytoplankton biomass was higher at downwind than at upwind nearshore sites, and the magnitude of this difference increased linearly with increasing wind speed. However, contrary to the generally assumed downwind phytoplankton aggregations, these differences were mostly due to upwelling activity and the dilution of phytoplankton at upwind nearshore sites. The distribution of individual taxa was also related to wind speed, but only during late stratification (except for cryptophytes), and these relationships were consistent with the buoyancy and motility of each group. On windy days, large diatoms and cyanobacteria concentrated upwind, neutrally buoyant taxa (green algae, small diatoms) were homogeneously distributed, and motile taxa (cryptophytes, chrysophytes, dinoflagellates) concentrated downwind. Predictable differences in the biomass and composition of phytoplankton communities could affect the efficiency of trophic transfers in nearshore areas.