1. science
2. weather

Reprint from
Aqua Fennica 22,2
Timo Huttula, Anu Peltonen, Amer Bilaletdin and
Matti Saura
THE EFFECTS OF CLIMATIC CHANGE ON LAKE
ICE AND WATER TEMPERATURE
l
1992
129
THE EFFECTS OF CLIMATIC CHANGE ON LAKE
ICE AND WATER TEMPERATURE
Timo Huttula, Anu Peltonen, Amer Bilaletdin and
Matti Saura
Huttula, T., Peltonen, A., Bilaletdin, A. & Saura, M. 1992. The effects of
climatic change on lake ice and water temperature. Aqua Fennica 22,2:129142.
The effect of climatic change on lake ice and water temperature was predicted
with a one dimensional vertical model. Applications were done in three Finnish
lakes of different sizes. Using the predictions from a climatological model and
from a watershed model the input data for the lake model was modified. The
results indicated about a two weeks delay in the first freezing of the lake and
about a two months shorter ice cover period than in the present climate. Small
sheltered lakes, having no spring turnover now, will have it in a changed
climate. The thermal stratification will be considerably steeper and the thermocline will be 5-8 m higher than in the present climate and the surface water
temperature will increase by about 5-6 “C. The water temperature in hypolimnion will be about 2-3 “C colder than in the present climate.
Index words: climatic change, ice cover, water temperature, model, lake.
Time Huttula, Anu Peltonen, Amer Bilaletdin and Matti Saura, Water and
Environment District of Tampere, PB 297, 33 101 Tampere, Finland.
INTRODUCTION
Changes in the composition of the atmosphere
and consequent global climate change will have
a profound impact on ice cover, water balance
and the stratification of lakes. These in turn may
change lacustrine ecosystems.
The effects of climatic change on terrestrial
and lacustrine ecosystems are studied widely within the Finnish Global Climatic Change Program (SILMU).
The aim of this paper is to discuss the effects of
climatic change on lake ice and water temperature. The work is a part of SILMU. A onedimensional vertical model for lakes has been
applied in three lakes of different sizes. Using
the predictions from a climatological model and
from a watershed model the input data for the
lake model was changed. The results are discussed in this paper.
THE MAIN EQUATIONS IN THE
MODEL
The PROBE-model (Svensson 1978), is used as
the hydrodynamical lake model of the project. A
simple water quality model has also recently
been incorporated into PROBE (Malve et al.
1991).
PROBE is a one-dimensional vertical model,
in which the water is assumed to be horizontally
homogenous. The model uses meteorological
data, water flow and material fluxes as input
data.
The basic PROBE-model calculates the development of the seasonal thermocline, water balance and the vertical distribution of eddy diffusivity. It is possible to include 14 state variables
for water quality modelling.
The basic assumptions in the model are that
the lake is horizontally homogenous and that
gravitational effects are assumed to obey the
Boussinesq approximation. The effects of the
130
earth’s rotation are included in the mean flow
equations, and the vertical exchange coefficient
is calculated using a two-equation model for
turbulence. A complete description of the model
and the numerical scheme was completed bvd
Svensson (1986).
where
The temperature equation in the model is:
At the bottom, a zero flux condition is used for
the temperature equation. For the momentum
equations on the surface quadratic wind stress
components are used with a drag coefficient of
1.3 10m3. At the lower boundary a zero velocity
is used for the momentum equations. The vertical mixing is modelled using a two-equation
turbulence model, the so-called k- &model. A
detailed description of the derivation and application of this model is given by Rodi (1980).
The density of the water is approximated by a
quadratic relation where it depends on the water
temperature.
The equations are solved in the finite difference form as integrated forward in time using a
implicit scheme and a standard tridiagonal matrix algorithm (Svensson 1978).
(1)
F, = the sum of the net long wave radiation, sensible hea flux and latent heat
flw
P = the water density and
c p = the specific heat of water (Table 1).
l
where
= time,
= water temperature,
= the vertical coordinate
(positive upwards),
VT = kinematic eddy viscosity,
k
2
0~ = turbulent Prandtl number and
ST = source/sink term.
The incoming short wave radiation is included
as a source term, as it penetrates the water surface and decays exponentially with depth. The
source term may also include the vertical advection term. The surface boundary condition for
the heat during the open water period is:
(2)
THE NET HEAT FLUX, F,
In the present work, the heat flux calculations
have been carried out as described by Sahlberg
(1983) and the following is mostly a quotation
by him. The fluxes controlling the heat content
of a lake through the air/water interface are
(Fig. 1):
Table 1. Constant values in the heat flux calculation PROBE-model. Sahlberg (1984).
Constant
g
a
PO
C
f
Sorn
E’
0
C
b
d
Pa
Cc
L
Ce
P
K
Li
Tk
Value
gravitational
acceleration
constant in the equation of state
maximum water density
specific heat of water
temperature of maximum density
solar constant
emissivity of a water surface
Stefan Bolzman’s constant
constant in the long wave radiation eq.
constant in the long wave radiation eq.
constant in the long wave radiation eq.
air density
sensible heat transfer coefficient
latent heat of evaporation
moisture transfer coefficient
constant in the latent heat equation
constant in the latent heat equation
latent heat of melting
absolute
temperature
9.81
8.25 lOA
999.975
4200
3.98
1395
0.97
5.67 1o-8
0.67
0.05
0.25
1.3
1.41 1o-3
2.5 lo6
1.32 1O-3
4370
5418
3.35 - 10’
273.15
l
l
Unit
ms - 2
Oc*
kg rn;’
J$tW rnB2
W-m-‘k4
mb-l /2
kg mm3
l
l
J ki- 1
-
l
-
JK m3
K
J kg- 1
K
131
T, = 1.041 -0.16(sec
z)~‘~
(5)
The amount of clouds and their height also affects short wave radiation. The “cloud function”
Ti is
Fig. 1. The heat fluxes controlling the heat contents of
the lake (Sahlberg 1984).
- net short wave radiation, F,
-
net long wave radiation F,
sensible heat flux, F,
latent heat flux, F,
precipitation, F,.
There are two more fluxes that affect the heat
content:
- heat flux, due to inlet/outlet from rivers and/
or groundwater, F,
- heat flux to or from sediments, Fb.
Short wave radiation, F s
The magnitude of short wave radiation (F,) penetrating a surface depends on:
- zenith angle of the sun,
- albedo of the surface,
- amount of clouds and water vapour in the
atmosphere.
The formulation of the short wave radiation
through the water surface is, according to Bodin
(1979):
3
FS = (l- a)$-, cosz(T, -A,) n (l- Ni(-Ti)) (WVZ-~)
i=l
*.>
(3)
where a is the albedo, So the solar constant
(S,= 1395 W mm2) and z the zenith angle. The
absorption by the water vapour in the air, A,, is:
A, = O.O77(u seezp3
(4)
where u is the amount of water in the atmosphere and set z=(cos z)-‘.
Scattering in a cloudless atmosphere is given
by Kondratyev (1969) as a scattering transmission function
T LOW
=0.35-0.015
TM I D D L E
=0.45-0.01
THIGH
=0.9 -0.04 set 2
sea
set 2
Ni is the amount of clouds of the different categories (low, middle, high). However, in this study
N will represent the total amount of clouds, and, as
an average, they are assumed to be middle high
clouds. According to this assumption, the only
“cloud function” used here is the TMIDD&unCtion. The magnitude of F, varies a lot during the
year. During the autumn and in the beginning of
the winter F, is much smaller than, for example,
the net long wave radiation.
Net long wave radiation, F,
The net long wave radiation consists of two
parts, one from the water surface to the atmosphere (F, f ) and the other from the atmosphere to
the water surface (Fl 4 ).
F1 = Fl t 41 4
(wm-2)
(6)
FIT = ElaT,4
where E’ is the emissivity of a water surface, ois
Stefan Bolzman’s constant and T, is the water
surface temperature (K).
The major problem in determining F1 is to get
a proper estimation of F1 J . Sahlberg (1983)
recommends a formulation which follows from
Brunt’s formula modified with a cloud factor of
F1J=oT,4(c+b&y)
(l+dN)
(7)
where T, is the air temperature, c, b, and d are
constants, e is the atmospheric water vapour
pressure and N is the cloud coverage.
The net absorbed atmospheric radiation is computed based on a 97 % absorptivity of the water
surface.
The net long wave radiation is one of the
major heat fluxes during cooling of a water mass.
F, is of the order lo2 W rne2.
132
The short wave radiation through ice
and snow
Sensible heat flux, F,
The formulation and the values of constants are
taken from Friehe & Schmitt (1976). They used
a bulk aerodynamic formula:
F, = P,c,~(C,, - cc2Ks - V)
(wm-2>
(8)
where pa is the air density, C, the specified heat
of water, Cc1 and-Cc2 are sensible heat transfer
coefficients and U is the wind velocity at 10 m.
The values of sensible heat transfer coefficients
depend on air stability, S,:
st =U(T,-Ta)
Latent heat flux, F,
The latent heat flux is also taken from Friehe and
Schmitt (1976) using a bulk aerodynamic formula:
(WK2)
(lo)
where L is the latent heat of evaporation, C, is
the moisture transfer coefficient and Qw and Qa
are the water vapour densities close to the water
surface and in the atmosphere respectively.
Heat fluxes from rivers and qroundwater, F,,
are of minor importance in the cases when the
retention time of the lake water is in the order of
years. The sediment heat flux, F,,, has to be
included in very shallow lakes in the winter time
simulations.
Thus the boundary condition for the model F,
can be written
FN=F1+Fc+Fe
F* = F (1 a ) i ,-q4-0.1)
s
s
- i 0
where F, = insolation falling on the upper layer
of the ice,
F: = the amount of insolation reaching
the ice-water, interface,
ai = albedo of the ice,
Ki = mean extinction coefficient of the
ice and
hi = ice thickness.
The variable i0 is the measure for the sun radiation which penetrates the upper 0.1 m of the ice.
The effect of snow cover is modelled with same
the penetration formulation as used for the ice
(Sahlberg 1988a and Sahlberg 1988b). The penetration of the radiation through the snow is
dependent on the albedo and extinction coefficient of snow.
In the model, ice is formed when the temperature of the upper 0.2 m top layer is less than zero
degrees. The growth of the ice is calculated with
a degree day method following Bengtsson and
Eneris (1977). For the melting of the ice, the
formulation from Ashton (1983) is followed.
The decreasing of the ice thickness ( A hi) is a
linear function of the air temperature. The final
destruction of the ice happens when the thickness of the ice is less than 10 cm and the wind
speed is greater than 6 ms-‘.
Ice growth
(11)
F, is so dominant during the spring and summer
that it is added as a source term in the temperature equation (1). During cooling in the autumn
F, has minor importance and it is incorporated in
the boundary condition.
if hi > O.lm
(12)
(9)
In stable conditions (S, < 0) &=0.0026 and
C,,=O.86E? In unstable conditions (0 < S, < 0)
C,1=0.002 and C,z=0.97E‘3 and in very unstable
conditions (S, > 25) C1=O.O and CcZ=1.46Ee3.
During cooling of a water mass, F, is one of the
major heat fluxes and varies between O-100 W
m- .
Fe =LC,u(Q,-12,)
In the winter time, sun radiation penetration to
lake water is strongly dependent on the albedo of
the ice and the extinction coefficient of the ice,
Ki. In the model, the results of Grenfell and
Maygut (1977) are applied (Sahlberg 1988a, Sahlberg 1988b). They found that in the first 0.1 m
of the ice, long wave radiation is absorbed and
short wave radiation decays exponentially from a
0.1 m depth of ice to the water surface:
03)
Where Kg has a value of 0.02 and Ta is the daily
mean air temperature.
Ice melting
133
Ahi = K,T, (m) if Ta >O
(14)
Where K,=4.3 10m3.
l
The model calculates ice growth and melting
based on the above equations and the direction
of the net surface heat flux, F,. Growth occurs
when the heat flux is directed from the ice surface towards the atmosphere and melting occurs
otherwise.
In the case of ice, the boundary conditions for
the momentum and temperature equations are
put to zero value at the upper boundary.
.
RESEARCH LAKES AND
CALIBRATION DATA
Lakes
,0
The PROBE-model has, as now, been applied in
six lakes with different catchment characteristics: Lappajarvi and the last basin in the Langelmavesi watercourse, influenced by agriculture as
well as wastewater loadings; Villikkalanjarvi and
Pyhajarvi, which are mainly affected by agriculture; Kalliojarvi, which is impacted by forestry
measures; and the pristine Hietajarvi. These lakes are located in southern and eastern parts of
the country.
Predictions of climate change effects have up
till now been conducted on lakes Lappajarvi,
Langelmavesi and Kalliojarvi (Fig. 2). Some physical characteristics of the lake basins are shown
in Table 2. The lakes have quite different sizes.
Lake Kalliojgrvi is the smallest and also quite
shallow. It is sheltered by the hills on the eastern
shore. Lappajarvi is a very big lake and has a
theoretical retention time of 2.5 years, whereas
the retention time for Lake Langelmavesi is only
60 days. The retention time for Lake Kalliojarvi
is 13 months.
Available calibration data
The available data is best for Lake Lappajarvi.
Surface temperature and water level in the lake
are observed daily during the open water period
by the Hydrological Office. The ice and snow
thickness is measured three times each winter
month. Ice formation and the destruction of the
ice was also observed. In the summer of 1988, an
automatic thermistor chain recorded water temperature in the lake for about two months.
For lake Lappajarvi, synoptic meteorological
data from Kauhava airport (about 30 km west)
was used. For Langelmavesi, the data was obtained from Tampere airport, which is about 20
km to the west of Lake Langelmavesi. For Kalliojarvi weather data was obtained from Kuorevesi airport, which is situated about 16 km east
from the lake. The model read the wind data in
three hours intervals, humidity and cloud data
were updated every 6 hours and daily mean values were used for air temperature.
The snow and ice thickness for Kalliojarvi
were observed five times during the winter and,
for Langelmavesi, the data was obtained from
the nearby outlet of the lake in Kaivanto.
The daily inflows and outflows were available
for each lake.
When water samples were also taken, water
temperature was measured on each lake every
week during the summer.
Hydrological conditions during calib.
ratlon years
In the following, the weather and runoff conditions during the period 1986-91 is described by
quoting the monthly reports of the Hydrological
Office. The monthly values of temperature and
wind speed in Kauhava are in Fig. 3. The outflow
of Lake Lappajarvi is in Fig. 4.
In 1986, spring began rather early, snow began to melt about two weeks earlier than normal, as did the ice cover over the watercourses.
The summer was warm and dry. The autumn
was exceptional ly wet and during the first weeks
also unusually cold. The end of the autumn
season was warmer than normal, with the mean
November temperature as much as 3°C above
the normal mean. Due both to the late onset of
winter and a November rainfall about double
the monthly mean , very high runoff values to the
watercourses were recorded in many areas. Lake
water levels were generally 50- 100 cm above the
seasonal mean in late November and early December and flow rates were as high as two- or
threefold compared with the seasonal mean flow
rates. Water resources were abundant in November-December.
In January 1987 the weather was dominated
by extreme frosts, which in southern Finland
were the heaviest recorded in the twentieth century. Ice covers grew considerably. The spring
runoff was lighter than normal. The summer
season was one of the coldest and and wettest of
134
0
41
2
’
4
”
6km
Lake
Lappa jarvi
KA
Lake
Kallio jarvi
Lake
Langelmhesi
Application area of PROBE in Lake Langelm%
vesi
Fig. 2.
Table 2. Physical characteristics of the lake basins
Lake
Area
km*
Volume
1O’m’
Mean depth
m
Max depth
m
KalliojCvi
LSngelmZvesi
LappajZrvi
0.3
11.2
161
1.1
95.0
1125
4.4
8.3
7.4
12.0
41.0
38.0
I
I
I
1986
I
1988
1987
I
1989
1990
1991
1990
1991
Years
-f)- m e a n
1987
1988
Years
1989
--E- Mean
--A- Maximum
--Q- Minimum
d ‘c
20
a
L
3
u
L
Q)
15
E”
a,
I--
0
10
m maximum
FZZZZI mean
5
I minimum
-5
-10
-15
1
2
3
4
5
6
7
8
9
10
11
12
Month
Fig. 3.
Wind speed and air temperature at Kauhava airport
1986-91
and temperature in a normal year.
136
m3 s
0
I
1986
I
1987
I
1988
I
1989
I
1990
1991
Years
--A-- Maximum
-Q-- Mean
--@- Minimum
Fig. 4. Outflow from Lake Lappajgrvi at Hanhikoski power station in 1986-91.
the century : precipitation levels were high, evaporation was unusually low and water temperatures remained well below average. At the beginning of August, heavy rainfall caused recordbreaking runoff volumes into watercourses. October and November were drier than normal. By
the end of the year water levels were generally
decreasing rapidly.
In 1988, melting of the thick snow cover coupled with heavy rainfall raised water levels in
spring and early summer to record heights for
the current century. June and July were exceptionally warm. Surface temperatures generally
reached 26°C and more in late June and July,
setting new all-time records for Finland.- The
autumn season began with near average conditions, but November was colder and frostier than
usual and winter came about a month earlier
than normal and water levels began to fall.
In the beginning of 1989 water levels were
high but very low by the end of the year. The
mean annual temperature was exceptionally high.
The winter was the warmest on record. Snow
began to melt early and the spring high water
occurred about two months earlier than normal.
The summer was long, warm and dry. As a
result, water levels decreased rapidly and were
very low in late summer and autumn and still
below normal level at the end of the year.
Both the weather and the hydrological conditions in 1990 were similar to those of 1989.
High levels of precipitation in the first months
of the year occurred both as snow and as rain.
Melting of snow began several weeks earlier than
normally. Both spring floods and the break-up
of lake ice were earlier than ever before on record. The summer was long and dry, although
not quite as warm as the two preceding summer
seasons. Water resources decreased during 1990
and were at a very low level by the end of the
year.
The recent trend of unusually warm years continued in 1991. Water reserves were low at the
beginning of year. The snow cover was thinner
than normal and melting of snow occurred early.
Sudden high meltwater runoff peaks occurred in
small watercourses but in large rivers the spring
flood was lower than normal. Heavy rainfall occurred during spring and again during the autumn, as a result of which water resources were
abundant by the end of the year. Water temperatures in August-September were expectionally
high, and freezing of the watercourses was several weeks late towards the end of the year.
r’
,
PREDICTIONS
The calibration of the model
The models were first calibrated against the lake
water temperatures during the open water period. The main parameters for the calibration were
the extinction coefficient and the wind. In the
137
Table 3. The values of calibration coefficient
Wind
in reducing wind at Lake Kalliojgrvi.
Calibration coefflent,
direction
c
310”-20”
20”-130”
1 SO”-200”
ZOO”-3 10”
4 ms
-1
1.0
0.8
1.0
0.8
case of Lappajarvi the wind velocity was somewhat increased from the velocities observed at
the airport (Malve et. al, 1991). Because Lake
Kalliojarvi is small and narrow (Fig. 2), the wind
calibration was used reducing the wind velocities especially in the direction of the transversal
axis of the lake (Table 3). The need to reduce
wind stress and solar radiation, because of hills
on the shoreline of the small lakes has been
discussed by Sahlberg (1988b). He suggested an
area dependent wind stress reduction on small
lakes. This method has not been tested in Finnish applications, since the calibration method
has proved to be acceptable.
In Fig. 5, the calculated and observed surface
temperatures of Lake Lappajarvi are seen. The
warming and cooling of the lake surface temperature is simulated well. The maximum values
differ due to the local heating of surface water at
the shallow observation site near the shore, whereas the model calculates the mean value of the
surface temperature in the whole lake. The vertical temperature profiles were also calculated well.
Fig. 6 is an example from Lake Lappajarvi in
summer 1988. The greatest deviations between
calculated and observed temperature occur in
late August when the model calculates the stra-
29
Oc
24
aJ
3
t 19
m
kl
g
t"
14
9
0
120
240
360
480
600
720
8 4 0 960 10801200
Date
Fig. 5. The surface temperature simulation in Lake Lappajgrvi (01 Ott 1986 30 Apr 1989)
when wind velocity
>
4 ms - 1
0.5
0.4
0.7
0.4
t&cation to be about 10 days longer than was
observed.
The calibration was done further using ice
data. The model was very accurate in calculating
the formation and destruction dates of the ice on
Lake Lappajarvi. The mean error in Lappajarvi
was 1.5 days (Malve et al. 1991). The ice thickness calculation deviated maximally about 10
cm and this was during the mild winter of 198889, when the snow melted several times on the
ice (Fig. 7). The snow observations turned out to
be too few for the model to calculate the heat
flux correctly during the period.
Simulation with a changed climate
The climatological predictions in the study are
based on the results from the GISS (Goddard
Institute for Space Studies) - model. The predictions of this model were the best choice of The
European Workshop on Interrelated Bioclimatic
and Land Use Changes in 1987 for simulations
of climatic change (Bach 1989).
Vehvilginen and Lohvansuu (1991) have used
a watershed model to predict the effects of climatic change on river discharges and snow cover
in Finland. Their work was based on the seasonal
temperature and precipitation predictions obtained from the GISS model when a doubling of
carbon dioxide relative to present day values was
assumed (the 2xC02 scenario). The temperature
and precipitation predictions of the GISS-model
for Finland are in Tables 4 and 5. For the lake
model, the input data of snow thickness and
water flow were obtained from their calculations.
In the present lake model applications meteorological data (air temperature, humidity, cloudiness and wind speed) from the years 19861991 were used. Air temperature was changed
according to the prediction of the GISS-model.
Predictions of changes in cloudiness, humidity
I
138
-----
Simulated
*+ Observed temp. in station N
temp.
-*-
Temperature
0
5 10 1 5 202530°c
0
5 10 1 5 2 0 2530 OC
0
5
5
10
10
15
20
n 25
25 i
cl
’I
I
I
0
,
I,
4‘I
, I
I
It
30
30
35
35 I
40 h
21Jun
m
0
5 I-
5
10 c
10
15
15
20
20
25
25
35
0 5 1015 202530 Oc 0 5 1015 20 25 30°C
0
r5 5 t
5
10 L
10
15 -
15
15
o 25 -
25
30
35
t
I
30 c
t
35
1
40 I
m
16Aug
I
20 -
i
1
//
25 -
35 -
40 L
m
24Aug
1.0
winter
1986-87
$9
0.8
z 0.7
0 5 10 15 202530 OC
T
5
IO
15
40 I
5Sept
m
Fig. 6. The water temperature simulation in Lake Lappajgrvi
12Jul
Or--
#9
I
I3
30 -
I iII
i
t
40
m
0 5 10 15 2025 30 "C
0
10 -
20
35
t
40 I
28Jun
m
OF--
Jr
5 20'-
0 5 10 1520 2530 "c
30
I%
40;
25May
m
5 IO 15 2 0 2 5 3 0 Oc
0
,
4
I
15
1=
5 20
Observed temp. in station S
I
t
40 I
m
22Sept
in summer 1988.
-Simulated
AObserved
winter
1987-88
c” 0.6
winter
1988-89
Y
.-w 0.5
z 0.4
fj 0.3
0.2
0.1
0.0
Fig. 7.
’
I
0
120
I
II
I
I
240 360 480 600
Date
Simulated and observed ice thickness on Lake Lappajgrvi
AI
I
I
\I
/
720 840 960 1080 1200
(01 Ott 1986 - 30 Apr 1989).
139
Table 4. The change in the seasonal air temperature in Finland. The GISS 2 x CO2 scenario. (According to
Vehvilainen and Lohvansuu, 199 1).
The change in air temperature “C
Area
Winter
South-Finland
North-Finland
+5.7
+G.o
Spring
+4.2
+4.6
Summer
+2.2
+3.1
Autumn
+4.2
+5.2
Annual
+4.1
+4.7
Table 5. The change in the seasonal air precipitation in Finland. The GISS 2 x CO2 scenario. (According Vehvilainen
and Lohvansuu, 199 1).
The change in precipitation mm in a month
Area
South-Finland
North-Finland
Winter
Spring
Summer
+13
+24
+25
+30
+31
+ll
and wind were not available. For the cloudiness
and humidity it was assumed that their relative
seasonal change is the same as the the change in
precipitation data. For the wind measured values
were used. In the earlier work (Kauppi et al.
1992) PROBE predictions were done with no
climatic change in humidity and cloudiness. In
the future the climate generator developed within the SILMU-program will be used for producing the meteorological input data.
RESULTS
In Lake Lappajarvi, which is the greatest of all
the study lakes, simulations were done in a period extending over three winters (Fig. 8). The ice
cover period in Lappajarvi will be significantly
shorter in a 2xC02 than in the present winter
situation. With all three winter data the formation of the ice cover was delayed about two
weeks. The date of ice melting occurred l-2
months earlier. The greatest change occurred in
the winter of 1988-89.
During the winters of
1987-88 and 1988-89 several periods with open
water occured.
Also the simulations of the PROBE-model for
the other two lakes indicated that the ice-covered period will be about two months shorter
than in the present (lxCOz) winter situation.
The first freezing of the lakes will be delayed
(Fig. 8). During the winter time there will be
ice-free periods, especially on large lakes, and
the final ice break-up will occur one or two
months earlier than today.
Autumn
+20
+21
Annual
+261
+264
Because of the earlier ice-melting in the spring,
the wind will mix the lake, and hence the temperature of the hypolimnion will slightly decrease compared with the present situation. On
the other hand, there will be no oxygen depletion in the hypolimnion in spring, due to reaeration and mixing. In small sheltered lakes the turnover conditions may change entirely. E.g. in
Lake Kalliojarvi which normally has no spring
turnover, because the heat obtained due to shortwave radiation penetration through the snowfree
ice rather quickly causes a weak summer stratification still when the ice is present. (Figs. 8 and
10). In the new situation (the 2xC02 scenario)
with shorter ice cover period the lake would also
mix in spring.
In summer the thermal stratification of the
lakes will be steeper and the temperature of the
epilimnion will increase about 5-6 “C (Fig. 911). The summer stratification will start about
one month earlier than today. The biggest changes of the epilimnetic temperature will occur in
May and September, when the change in, e.g.,
Lake Lappajarvi will be more than 7.5 “C (Fig.
9). The thermocline will be about 5-8 m higher
than today and the hypolimnion will be about 23 “C colder. The longer stagnation period means, in practice, more serious problems with
oxygen depletion in the hypolimnion during the
late summer.
The results do not differ significantly from the
earlier results (Kauppi et. al 1992), where the
humidity and cloud data were not changed. In
the present work, this change was done assu-
14-Q
Lake Lappaj&vl
..
.
.
.
.
.
.
.
.
.
.
.
,
1986-87
.
i
1987-88
:
.
*
. .
i
.
lDa8-89
.
*
Lake Kallioj6rvl
.:
.
1988-89
*
.
.
.
.
.
:
;
.
.
lD89-90
.
.
.
:
.
L----1.10.
31.10
30.11.
30.12.
29.1.
28.2.
30.3.
29.4.
Lake Lhgelmhesi
.
.
.
.
;
1990-91
.
.
.
* lxCO2
.
.
.
*
.
..
..
..
.
.
.
l
.
t
.
.
I
- __. t30.12.
_..-+
--.----29.1.
------...
+
-+-- -----~ - ~.-----------+
28.2.
30.3.
Fig. 8. The ice cover in research lakes in present cilmate and in the 2 x CO, climate.
29.4.
29.5.
141
resulted in an increased temperature in the epilimnium. The difference in temperatures simulated with cloud and humidity data not changed
was maximally 2 “C.
DISCUSSION
0
120
240
360
480
600
720
840
960
1080
1200
1320
Date
Fig. 9. The surface water temperature of Lake Lappajgrvi in the present situation and in the 2 x CO, climate.
- IXCOZ
surface
1
---2xc0, I
I'\\, 1%
I I' \ - 1 x co2
b
, . . . . . . . . . . 2 x co2 bottom
t.:
1
I
$,
, v/4
L
iii
E
: 10
I
0 c
880622
I
880930
890108
890418
890727
Date
891104
900212
900523
1’
90 0831
Fig. 10. The surface water temperature of Lake KalliojZrvi in the present situation and in the 2 x CO, climate.
aJ
5
20
l-0
tJ
+I
E"
:
10
1
0
1
900212
900523
900831
901209
Date
910319
910627
911005
920113
Fig. 11. The surface water temperature of Lake L&gelmgvesi in the present situation and in the 2 x CO,
climate.
ming that humidity and cloudiness have the
same relational change as precipitation computed
by the GISS-model. The main differences occurred in the middle of the summer when increased
humidity and cloudiness increased the longwave
radiation and decreased the evaporation. These
In Finnish latitudes, a climatic change will have
a major effect on the snow and ice cover. This in
turn will have an effect on mixing and stratification.
The PROBE-model, with the heat flux calculations presented above, has proved to be very
good tool in calculating the vertical temperature
structure in a lake with a fairly open surface.
There is not much calibration needed and the
necessary data can be obtained from synoptic
weather stations and water temperature measurements from the water authorities.
Kuusisto (1989) presented results of the change of the ice cover period in Finland in the GISS
2xCOz-scenario. According to his results the ice
cover length will be 40-60 days shorter than in
the present climate. This corresponds well with
present results. Kuusisto based his calculations
on a statistical dependency between mean annual air temperature and the length of the ice
cover.
The simulation of clouds is a problem in global climatic models as Dickinson (1986) and
Bach (1989) have pointed out. The wide variety
of cloud types and the fact that most cloud
properties are under the grid scale of the present
climatic models make cloud prediction uncertain.
The sensitivity of the model to the change of
air temperature was studied in the application of
Lake Langelmavesi. Air temperature was assumed to increase one degree less compared to the
estimations of the GISS-model. In this case the
ice-cover period was only about one month shorter than in the present situation. Ice-free periods
did not occur in winter. Also the results of in
summer time became more comparable to those
of in present situation.
The water temperature of a lake is sensitive to
wind speed and solar radiation. Thus, the results
of the ice-cover period must be considered to be
more reliable than the summer results.
The effect of wind is the main driving force for
the vertical mixing in lakes. In the present work
the wind data was not changed. The last few
years have been exceptionally warm in Finland
and also more windy than before. If this means
that winds will also increase in the 2xC0, clima-
142
te, the stratification will be effected. This
lead to a more sharper thermocline than
dieted now and great variations in summer
face temperatures in some lakes, when the
hypolimnetic water is mixed with surface
ters.
may
presurcold
wa-
TIIVISTELMh;
Ilmaston muutoksen vaikutusta jarven jzhan ja
veden lampotilaan arvioitiin yksidimensioisen
vertikaalimallin avulla. Mallia sovellettiin kolmeen erikokoiseen suomalaiseen jarveen. Laskelmat perustuivat GISS-mallin lampotila- ja sadeskenaarioihin seka valuma-aluemallista saatuihin virtaama- ja lumenvesiarvoskenaarioihin. Pilvisyyden oletettin muuttuvan sadannan suhteessa. Tuuliarvoja ei muutettu.
Tulokset osoittavat jarvien ensijztymisen tapahtuvan kaksi viikkoa myohemmin ja jtipeitteen kestoajan olevan noin kaksi kuukautta lyhyemman kuin nykyilmastossa. Pienilla jarvilla,
joilta kevattayskierto puuttuu nykyilmastossa,
tapahtuisi sekoittumista kevaisin muuttuneessa
ilmastossa.
Jarvien lampotilakerrostuneisuus tulisi olemaan
huomattavasti jyrkempi ja harppauskerros jti
noin 5-8 m ylemmaksi kuin nyky-ilmastossa.
PZllysveden lampotila nousee keskimzrin 5-6
“C. Alusveden lampotila j% noin 2-3 “C kylmemmaksi k u i n nykyzn. Kesan kerrostuneisuuskauden pituuden kasvu 1isZ alusveden happiongelmia.
Mallin antamat tulokset ovat luotettavimpia
talven jzpeitteen laskennan ostalta. Tuuliennusteiden puuttumisen vuoksi kesan kerrostuneisuuden laskentaan liitty epavarmuustekijoita. Mikali tuulien voimakkuus kasvaa avovesikaudella, jarvien kerrostuneisuus on nyt laskettua jyrkempi. Toisaalta alusveden kumpuaminen yleistyy joillakin jarvilla ja sen vaikutukset
jarvien biologiaan voivat olla varsin merkittavat.
REFERENCES
Ashton, G. D. 1983. Lake ice decay. Cold Regions Science and Technology, 8: 83-86.
Bach, W. 1989. Projected climatic changes and impacts
in Europe due to increased C02. In: Conference on
Climate and Water. Helsinki, Finland. ISBN 95 l861-668-X. pp. 31-50.
Bengtsson, L. & Eneris, E. 1977. Istillv& i sjijar analyserad med graddagarsmetoden. TULEA 1977: 09,
Lulei, S w e d e n .
Bodin, S. 1979. A predictive numerical model of the
Atmospheric Boundary Layer based on the Turbulent
Energy Equation. SMHI-rapporter, RMK 13 (1979).
Dickinson, R.E. 1986. How will climate change? The
climate sytem and modelling of fututre climate. In:
B . B o l i n , B.R.DoBs, J.Jager a n d R . A . Warrick (eds.),
The Greenshouse effect, Climaticla Change and
Ecosvstems. 206-270. Wilev. Chichester.
Friehe, C A, and Schmitt: K F.‘l976. Parameterization
of air-sea interface fluxes of sensible heat and moisture by bulk aerodynamic formulas. Journal of Physicai Oceanography, 6(6).
Grenfell, T. C. & Maygut,
G. A. 1977. The optical
properties of ice and snow in the arctic basin. Glaciol., 18 (80).
Kauppi, L., Frisk, T., Forsius, M., Huttula, T., Posch,
M., Bilaletdin, A., Kamari, J., Kallio, K., Peltonen,
A. 8~ Saura, M. 1992. Effects of Climatic Change,
Air Pollutants and Land Use on Lake Ecosystems. In:
Kanninen, M. & Anttila, P. (eds.). The Finnish Research Programme on Climate Change. Helsinki,
Finland. ISBN 951-37-0846-2.
pp. 126-131.
Kondratyev, K. 1969. Radiation in the atmosphere. International Geophysics Series, Vol. 12, Academic
Press, New York.
Kuusisto, E. 1989. Snow and ice - nonrenewable natural
resources in the future. In: Conference on Climate
and Water. Helsinki, Finland. ISBN 95 l-861-668X. pp. 300-318.
Malve, O., Huttula T. & Lehtinen K. 1991. Modelling
of eutrophication and oxygen depletion in Lake Lappajarvi.
In: “Firs’t Int. Conf. on Water Pollution:
Modelling, Measuring and Prediction” Southamptonsept. 1991. ISBN 1-85166-697-4~~.
111-124.
Wrobel, L.C & Brebbia, CA. (eds.). Water Pollution:
Modelling, Measuring and Prediction: 11 l-l 24.
Computational Mechanics Publications, Southampton . ASBN 1-85166-697-4.
Rodi, W. 1980. Turbulence models and their applicatio n i n h y d r a u l i c s - a state of the art review. Int. Assoc.
Hydr. Res. (IAHR), Delft, The Netherlands.
Sahlberg , J. 1983. A hydrodynamical model for calculating the vertical temperature profile in lakes during
cooling. Nordic Hydrol. 1983: 239-254.
Sahlberg, J. 1988a. Modelling the thermal regime of a
lake during the winter season. Cold Regions Technology, 15. Elsevier Publishing Company, Netherlands. pp. 151-159.
Sahlberg, J. 1988b. Beragning
av vinteruppvarming och
t e r m o k l i n u t v e c l i n g i T u l e b o s jiin. Rapport R79: 1988.
Byggforskningsrgdet.
Stockholm. ISBN 91-5404948-2.
Svensson, U. 1978. A mathematical model for the seasonal thermocline. Ph.D. dissertation, Report 1002,
Dep. Water Resour., Univ. of Lund, Sweden. 187
PP.
Svensson, U. 1986. PROBE - An instruction manual.
Rep.Oceanofr. 10, Swedish Meteorological and Hydrological Institute (SMHI). S-60 176, Sweden.
Vehvilginen
B. & Lohvansuu J. 1991. The effects of
climate change on discharges and snow cover in
Finland. Hydrological Sciences-Journal 36: 109- 12 1.
Dt)tp & G-W%’
Received 2.3 November 1992
Accepted 3 1 March 1993
LEGI+ m=
~w1b-E
@+sl r/u
fir=@%’
*&
7r4R~p~r~/ bomm-