parameterization of lakes in nwp and climate models - hzgsurface albedo (mironov and ritter 2003,...

COSMO-CLM-ART Training, Langen, Germany, 23 – 31 March 2015

Parameterization of Lakes in NWP and Climate Models

Dmitrii V. Mironov

German Weather Service, Offenbach am Main, Germany

Hermann Asensio, Erdmann Heise, Ekaterina Machulskaya, Bodo Ritter(German Weather Service, Offenbach am Main, Germany) Sergey Golosov(Institute for Lake Research, Russian Academy of Sciences, St. Petersburg, Russia) Georgy Kirillin (Leibniz Institute of Freshwater Ecology and Inland Fisheries, Berlin, Germany) Ekaterina Kourzeneva(Finnish Meteorological Institute, Helsinki, Finland, and Russian State Hydrometeorological University, St. Petersburg, Russia) Arkady Terzhevik(Northern Water Problems Institute, Russian Academy of Sciences, Petrozavodsk, Russia)

Outline

• Parameterization of lakes in NWP and climate models – the problem

• The lake parameterization scheme “FLake” • FLake in the COSMO model • COSMO-FLake performance • Conclusions and outlook

Parameterization of Lakes in NWP and Climate Models A Twofold Problem

(1a) The interaction of the atmosphere with the underlying surface is stronglydependent on thesurface temperature and its time-rate-of-change. (Most) NWPmodels assume that the water surface temperature can be keptconstant overthe entire forecast period. The assumption is doubtful for small-to-medium sizerelatively shallow lakes, where the diurnal variations of the surfacetemperature reach several degrees. A large number of such lakes will becomeresolved-scale features as the horizontal resolution is increased.

(1b) Apart from forecasting the lake surface temperature, itsinitialization isalso an issue.

(2) Lakes strongly modify thestructure and the transport properties of theatmospheric surface layer. A major outstanding question is theparameterization of the water-surface roughness with respect to wind (e.g.limited fetch) and to scalar quantities.

Lake Regions: Finland, Karelia

Lake Regions: Khanty-Mansiisk region

(middle Ob’ river)

Lake Regions: Canada

Lake Parameterization Schemes for NWP and Climate Models

(e.g. Croley 1989, 1992, Croley and Assel 1994, Hostetler and Bartlein 1990, Hostetler 1991, Hostetler et al. 1993, 1994, Barrette and Laprise 2005, Bates et al. 1993, 1995, Ljungemir et al. 1996, Goyette et al. 2000, Tsuang et al. 2001, Song et al.

2004, León et al. 2005, 2007, Long et al. 2007, Mackay 2005, Mackay et al. 2009, Stepanenko and Lykosov 2005, Stepanenko 2007, Stepanenko et al. 2010, Subin et al. 2012)

• One-layer models, complete mixing down to the bottom Neglect stratification ⇒ large errors in the surface temperature

• Turbulence closure models, multi-layer (finite-difference) Describe the lake thermocline better ⇒ expensive computationally

A compromise betweenphysical realism and computational economy

is required

A two layer-model with a parameterized vertical temperature structure

The Concept

Put forward by Kitaigorodskii and Miropolsky (1970) to describethe temperature structure of the oceanic seasonal thermocline.The essence of the concept is that the temperature profile in thethermocline can be fairly accurately parameterized through a“universal” function of dimensionless depth, using thetemperature difference across the thermocline, ∆θ=θs-θb, anditsthickness, ∆h, asappropriate scalesof temperature and depth:

.)(

)(),(

)(

),()(

th

thz

t

tzts∆−==

∆− ςςϑθ

θθ

Analogy with the Mixed -Layer ConceptUsing θs(t) and h(t) as appropriate scalesof temperature and depth, thetemperature profile in the upper mixed layer is represented as

.)(

),()(

),(

th

z

t

tz

s

=Φ= ξξθ

θ

Since the layer is well mixed, the “universal” functionΦ(ξ) is simply a constant equal to1.Then, integrating the heat transfer equation (partial differential equation inz, t)

z

w

t ∂′′∂−=

∂∂ θθ

over z from 0 to h(t), reduces the problem to the solution of anordinarydifferential equationfor θs(t),

h

hQQ

dt

d ss )(−=θ

The Lake Model “FLake”

• the mean temperature of the water column,• the surface temperature,• the bottom temperature,• the mixed-layer depth,• the shape factor with respect to the temperature profile inthe thermocline,• the depth within bottom sediments penetrated by the thermal wave, and• the temperature at that depth.

The model is based on the idea of self-similarity (assumed shape) of the evolvingtemperature profile. That is, instead of solvingpartial differential equations(in z, t) forthe temperature and turbulence quantities (e.g. TKE), the problems is reduced tosolving ordinary differential equationsfor time-dependentparameters (variables) thatspecify the temperature profile. These are (optional, modules can be switched off)

In case of ice-covered lake, additional prognostic variables are • the ice depth, • the temperature at the ice upper surface, • the snow depth, and the temperature at the snow upper surface.

Important! The model does not require (re-)tuning.

(a) The evolving temperature profile is characterised by several time-dependent variables, namely, thetemperatureθs(t) of the mixed layer, its depthh(t), the bottom temperatureθb(t), and thetemperature-profile shape factorCθ(t). Optionally, the depthH(t) within bottom sedimentspenetrated by the thermal wave and the temperatureθH(t) at that depth can be computed.

θs(t)θb(t)

(a)

θL

θH

(t)

h(t)

D

L

H(t)

Cθ(t)

Schematic representation of the evolving temperatur e profile

(b) For ice-covered lakes, additional variables are the temperatureθI(t) at theice upper surface and the ice thicknessHI(t), and (optionally) thetemperatureθS(t) at the snow upper surface and the snow thicknessHS(t).

θs(t) θb(t)θI(t)θS(t)

(b)θ

L

θH

(t)

h(t)

D

L

H(t)

-HI(t)

-HI(t)-H

S(t)

Snow

Ice

Water

Sediment

Cθ(t)

FLake in NWP and Climate Models: External Parameters

• geographical latitude(easy)

• lake fractionof the NWP model grid-box (not so easy)

• lake depth(not easy at all, lack of data, etc.)

• typicalwind fetch

• optical characteristics of lake water(extinction coefficients with

respect to solar radiation)

• depth of the thermally active layer of bottomsediments, temperature

at that depth(cf. soil model parameters)

Default values of thelast fourparameters can be used.

Lake Fraction

Lake-fraction external-parameter field for the COSMO-EU numerical domain of the COSMO model (ca. 7 km horizontal mesh size).

Lake Depth

Lake depths for “Northern Europe” region of the COSMO-EU numerical domain (ca. 7 km horizontal mesh size).

Lake Depth

Lake depths based on COSMO-DE external-parameter field (ca. 2.8 km horizontal mesh size).

FLake in the COSMO Model: Configuration

• bottom sediment module is switched off (heat flux through the water-

bottomsediment interface is zero)

• snow above the lake ice is not considered explicitly, the effect of snow is

accounted for implicitly through the temperature dependence of the ice

surface albedo (Mironov and Ritter 2003, 2004, Mironov et al. 2012)

• turbulent fluxes at the surface are computed with the current COSMO-

model surface-layer scheme (Raschendorfer 2001); optionally, the new

surface-layer scheme (Mironov et al. 2003) can be used

• no tile approach: lakes are the COSMO-model grid-boxes with

FR_LAKE>0.5, otherwise land or sea water

• 2D fields of lake fraction and of lake depth based on data (Kourzeneva

2009, 2010, Kourzeneva et al. 2012), default values of other lake-specific

parameters

FLake in the COSMO Model: Parallel Experiment

• COSMO-model parallel experiment over one year, 1 January

through 31 December 2006, using the LM1 numerical

domain of DWD

• The entire COSMO-model data assimilation cycle except

that the lake surface temperature is not re-initialised through

the SST analysis but is predicted by FLake

• “artificial” initial conditions at the cold start, where the lake

surface temperature is equal to the COSMO-model SST

from the analysis

FLake in COSMO: Results from Parallel Experiment 56 32

1 January – 31 December 2006

Lake Hjälmaren, Sweden (mean depth = 6.1 m)• Black – lake surface temperature from the COSMO SST analysis • Green– lake surface temperature computed with FLake



Lake Balaton, Hungary (mean depth = 3.3 m)• Black – lake surface temperature from the COSMO SST analysis • Green– lake surface temperature computed with FLake



Lake Balaton, Hungary (mean depth = 3.3 m). Ice thickness computed with COSMO-FLake.

Ice melting: very beginning of March

Freeze-up: 10 January



Lough Neagth, UK (mean depth = 8.9 m)• Black – lake surface temperature from the COSMO SST analysis • Green– lake surface temperature computed with FLake

FLake within COSMO -EU/DE of DWD

Flake is used operationally at DWDsince 15 December 2010within COSMO -EU (ca. 7 km horizontal mesh size), andsince 18 April 2012 within COSMO-DE (ca. 2.8 km meshsize).

• Results of testing of FLake within COSMOare neutral to slightlypositive.

• Verification against observational data indicate an improvement ofsome scores such as 2m-temperature in regions where many lakesare present (e.g. Scandinavia).

• The use of FLake allows to avoid some unwanted situations, e.g. anartificial cold air outbreak. This may occur in winter when a lakethat is frozen in reality (lowsurface temperature) is treated as openwater (high surface temperature) within COSMOdue to theshortcomings of water surface temperature analysis scheme.

The performance of FLake within COSMO-EU/DE is monitored.

Conclusions and Outlook• FLake is implemented into the COSMO model (setllake=.FALSE. to run

without FLake)

• Since 15.12.2010/18.04.2012 Flake is operational at DWD within COSMO-EU/DE (ca. 7 kmand ca. 2.8 kmmesh size, respectively)

• Documentation and synopsis of FLake routines are ready

• FLake is operational at DWD within global NWP model ICON since 20January 2015; tiled surface scheme is currently used in ICON(effect of sub-grid scale lakes is accounted for)

• Monitor operational results

• Update external-parameter fields

• Account for SGS lakes within COSMO (tiled surface scheme, c/o EkaterinaMachulskaya and Jürgen Helmert)

• Long termprospective: explicit treatment of snow over lake ice, three-layertemperature profile, salinity, data on optical propertiesof lake water

External parameters and tiled surface scheme

The lake-fraction external-parameter field based on the lake-depth data fromKourzeneva (2010) and GlobCover physiographic data. The horizontal size of theCOSMO-model grid is ca. 7 km.

InfoFLake Web Pagehttp://lakemodel.net(mirror http://nwpi.krc.karelia.ru/flake),

c/o Georgiy Kirillin and Arkady Terzhevik

Online FLake versionat http://lakemodel.net(take a look and have fun!)

References

Kirillin, G., J. Hochschild, D.Mironov, A. Terzhevik, S.Golosov, and G. Nützmann, 2011: FLake-Global: Onlinelake model with worldwide coverage.Environ. Modell. Softw., 26, 683-684.

Kourzeneva, E., 2010: External data for lake parameterization in Numerical Weather Prediction and climatemodeling.Boreal Env. Res., 15, 165-177.

Kourzeneva, E., H. Asensio, E. Martin, and S. Faroux, 2012: Global gridded dataset of lake coverage and lakedepth for use in numerical weather prediction and climate modelling. Tellus A, 64, 15640.doi:10.3402/tellusa.v64i0.15640

Mironov, D. V., 2008: Parameterization of lakes in numerical weather prediction. Description of a lake model.COSMO Technical Report, No. 11, Deutscher Wetterdienst, Offenbach am Main, Germany, 41 pp.

Mironov, D., E. Heise, E. Kourzeneva, B. Ritter, N. Schneider, and A. Terzhevik, 2010: Implementation of thelake parameterisation scheme FLake into the numerical weather prediction model COSMO.Boreal Env. Res.,15, 218-230.

Mironov, D., B. Ritter, J.-P. Schulz, M. Buchhold, M. Lange,and E. Machulskaya, 2012: Parameterization of seaand lake ice in numerical weather prediction modelsof the German Weather Service. Accepted for publicationin Tellus A.

Further references athttp://lakemodel.net

FLake Applications

• Lake parameterization scheme for NWP and climate models

(computationally efficient, can be used to treat a large number of

lakes)

• Single-column lake model in a stand-alone mode (assessment of

response of lakes to climate variability, estimation of evaporation

from the water surface, aid in design of ponds and reservoirs, etc., a

cost-effective decision-making tool)

• Physical module in models of lake ecosystems (a sophisticated

physical module is not requiredbecause of large uncertainties in

chemistry and biology)

• Educational tool (simple but incorporates much of the essential

physics)

FLake in NWP and Climate Models

As a lake parameterization scheme, FLake is

• implemented, or on the way, into a number of NWP and

climate models (COSMO, ICON, HIRLAM, UK Met Office

Unified Model, NWP model suite of Meteo France, ECMWF

IFS, CLM, RCA, Canadian Regional Climate Model),

• used as a lake parameterization module in the surface

schemes TESSEL, SURFEX, and JULES,

• used operationally at the German Weather Service within

COSMO-UE/DE and ICON, used operationally at the Finish

Meteorological Institute within HIRLAM

Thank you for your attention!

Acknowledgements: Frank Beyrich, Michael Buchhold, Ulrich Damrath, Günther Doms, JochenFörstner, Helmut Frank, Thomas Hanisch, Jürgen Helmert, Peter Meyring, Van Tan Nguyen, UlrichSchättler, Christoph Schraff (DWD), Andrey Martynov (UQAM), Burkhardt Rockel (GKSS), PatrickSamuelsson (SMHI), Laura Rontu (FMI), Zachary Subin (UC Berkeley).

EU Commissions, Projects INTAS-01-2132 and INTAS-05-1000007-431; Nordic Research Boardthrough the Nordic Networks on Fine-Scale Atmospheric Modelling (NetFAM) and Towards Multi-Scale Modelling of the Atmospheric Environment (MUSCATEN).


Single-Column Tests

• Kossenblatter See, Germany (52 N, depth = 2 m)

• Lake Krasnoye, Russia (60 N, depth = 8 m)

• Lake Pääjärvi, Finland (61 N, depth = 15 m)

• Ryan Lake, USA(45 N, depth = 9 m)

Forcing in single-column mode Known from observations: • short-wave radiation flux, • long-wave radiation flux from the atmosphere.

Computed as part of the solution (depend on lake surface temperature):• long-wave downward radiation flux from the surface, • fluxes of momentum and of sensible and latent heat, • for ice-covered lakes, surface albedo.

Kossenblatter See, 8-21 June 1998.

Water surface temperature (θf is the fresh-water freezing point) • Dots- measured • Line - computed

0 48 96 144 192 240 288 33618

19

20

21

22

23

24

25

26

time, h

θs-

θ f ,

K


Friction velocity in the surface air layer

• Symbols- measured

• Line - computed

216 240 264 288 3120

0.1

0.2

0.3

0.4

0.5

time, h

u*

, m⋅ s

-1


• Sensible heat flux Qse• Latent heat flux Qla• Symbols– measured

• Lines– computed

216 240 264 288 312-80

-60

-40

-20

0

20

time, h

Qse

, W

⋅ m-2

216 240 264 288 312-300

-250

-200

-150

-100

-50

0

time, h

Qla

, W

⋅ m-2

Lake Krasnoye, 1 May - 31 October 1970.

Water surface temperature θs (θf is the fresh-water freezing point)Dots– measured, line - computed

120 150 180 210 240 270 3000

5

10

15

20

25

time, day

θs-

θ f ,

K

Lake Pääjärvi, 1 May 1999 - 31 August 2002.

Water surface temperature θs (θf is the fresh-water freezing point) Dots– measured, line - computed

120 240 360 480 600 720 840 960 1080 12000

5

10

15

20

25

time, day

θs-

θ f ,

K

ice ice ice

Lake Ryan, November 1989 – November 1990.

Surface temperature, mean temperature of the water column, bottom temperature

Dotted – measured, solid – modelled

-30 0 60 120 180 240 3000

5

10

15

20

25

30

35

time, day

θ- θ

f , K

Ice

? (sensor malfunction)

Lake Ryan, April – November 1990.

Mean temperature of the water column

Measuredvs. modelled

90 120 150 180 210 240 270 3000

5

10

15

20

time, day

θm

- θf ,

K

Lake Ryan, December 1989.

• Solid - modelled ice surface temperature • Dotted- temperature measured with the uppermost sensor

-30 -20 -10 0

-25

-20

-15

-10

-5

0

time, day

θi- θ

f , K

Single-Column Tests: Perpetual Year Solution

• Lake Swente, Latvia (56 N, depth = 17.5 m, transparent

waterγ=0.3 m-1)

• computed atmospheric radiation fluxes

• climatologically mean forcing (1961 – 1964)

• measured water temperature at a number of depths

• no flux measurements

Lake Swente, Perpetual Year.

Surface temperature, mean temperature of the water column, bottom temperature

Symbols – measured, lines – modelled

0 60 120 180 240 300 360

0

5

10

15

20

time, day

θ- θ

f , K

Ice

Lake Fraction

Lake-fraction external-parameter field for the LM1 numerical domain (DWD) of the COSMO model based on the GLCC data set (http://edcsns17.cr.usgs.gov/glcc/) with 30 arc sec resolution, that is ca. 1 km at the equator.

Lake Depth

Lake depths for the LM1 numerical domain of the COSMO model. The field is developed using various data sets. Each lake is characterised by its mean depth.



Lake Hjälmaren, Sweden (mean depth = 6.1 m)

Ice thickness (left) and ice surface temperature (right) computed with COSMO-FLake



Neisiedlersee, Austria-Hungary (mean depth = 0.8 m)• Black – lake surface temperature from the COSMO SST analysis • Green– lake surface temperature computed with FLake



Lake Vänern, Sweden (mean depth = 27 m)• Black – lake surface temperature from the COSMO SST analysis • Green– lake surface temperature computed with FLake


Parameterization of Lakes in NWP and Climate Models

Dmitrii V. Mironov

German Weather Service, Offenbach am Main, Germany

Hermann Asensio, Erdmann Heise, Ekaterina Machulskaya, Bodo Ritter(German Weather Service, Offenbach am Main, Germany) Sergey Golosov(Institute for Lake Research, Russian Academy of Sciences, St. Petersburg, Russia) Georgy Kirillin (Leibniz Institute of Freshwater Ecology and Inland Fisheries, Berlin, Germany) Ekaterina Kourzeneva(Finnish Meteorological Institute, Helsinki, Finland, and Russian State Hydrometeorological University, St. Petersburg, Russia) Natalia Schneider(University of Kiel, Kiel, Germany) Arkady Terzhevik(Northern Water Problems Institute, Russian Academy of Sciences, Petrozavodsk, Russia)

parameterization of lakes in nwp and climate models - hzgsurface albedo (mironov and ritter 2003,...

Documents