Impact of climate change on Latvian water environment

WP1: Impact of the climate change on runoff, nutrient fluxes and regime of Gulf of Riga

Uldis Bethers

Laboratory for mathematical modelling of environmental and technological processes Faculty of Physics and mathematics,University of Latvia

WP1. GOAL

Modelling of several scenarious of the change of water environment using the existing climate change scenarious for Baltic Sea region

WP1. TASKS

WP1a. Evaluate and adapt the results from the regional climate models, and design the series of data which form the state of the water objects. Scenarios

WP1b. Modeling of surface water and nutrients runoff for Latvia. Preparation of data series of river runoff for climate change scenarios

Calculation of data series of nutrient runoff to the Gulf of Riga

WP1c. Adapt 3D sea state models to produce the data series for the forecast of biogeochemical processes and sea ecosystem evolution.Oceanographic modellingWP1d. Provide modelling and data analysis support for other WPs. Support

Climate change scenarios (IPCC)

Regional climate models

Climate change scenarios adapted for Latvia

River runoff scenarios

Sea state scenarios (Gulf of Riga)

Impact assessment (on Latvian water environment)

Information flows

Global circulation models

Nutrient runoff scenarios

21 model (PRUDENCE)

118 observation stations and selected RCM mesh

Daily meteorological data (temperature, precipitation, wind, humidity, cloudiness)

Methodics for measurement and comparison of RCM skill for control period

Double downscaling: bias correction (statistical downscaling via histogram equalisation) of dynamically (via RCM) downscaled GCM data – T, p, r

Climate scenarios 1A

Histogram equalisation for moving time window [instead of daily or monthly or seasonal equalisation]

OBS

REF

A2

MODA2

MODB2

B2

MODREF

600

650

700

750

800

850

900

4 5 6 7 8 9 10 11

Temperatūra

No

kriš

ņi

Daily data series for Latvia – contemporary climate, climate change scenarios

Okt

S ep

Aug

J ūl

MayDec

MarF eb

J an

Okt

S ep

Aug

J ūl

J ūnNovDec

J an

Apr

Mar

F eb

0.5

1

1.5

2

2.5

3

-10 -5 0 5 10 15 20 25

Temperatūra, deg C

Nokrišņi, m

m/d

Mūsdienu klimats

S cenārijs A2

Insight : T-p diagram for Dobele, contemporary climate and A2 scenario

Winter onlyin Feb

Spring 2weeks earlier

Summer 2 monthslonger, hot, dry

Pleasant Sep/Oct

Rainy, longautumn

A2

B 2

1961-1990 R K M

1961-1990 novērojumi

1.50

1.55

1.60

1.65

1.70

1.75

1.80

5 6 7 8 9 10 11

T, deg C

p, m

m/d

Dobele: comparison scenarios vs. observations

A2

B 2

1961-1990 R K M

1961-1990 novērojumi

1978-2007

1.50

1.55

1.60

1.65

1.70

1.75

1.80

5 6 7 8 9 10 11

T , deg C

p, m

m/d

A2

B 2

Novērotās klimata izmaiņas

1961-1990 R K M

1961-1990 novērojumi

1978-2007

1.50

1.55

1.60

1.65

1.70

1.75

1.80

5 6 7 8 9 10 11

T , deg C

p, m

m/d

Noteces modelēšana 1B

• Hidroloģisko modeļu ansambļa pieeja 2008• Noteces kalibrācija / verifikācija ūdensobjektiem• Upju noteces scenāriji

Atziņa 2008 – upju notece RJL sateces baseinā samazināsies vismaz par 5-10%

• Reģionālā upju noteces izmaiņu analīze Latvijai 2009• UBA noteces scenāriji 2009• Biogēnu noteces scenāriji 2009 (paldies Bārbelei )

Atziņa – hidroloģiskā režīma daudzveidība Latvijā samazināsies

NOVITĀTE PASAULĒ

“Double ensemble forecast: ensemble of RCM vs. ensemble of hydrological models”

Regional climate modelRiver run-off

Hydrological model(independent from RCM)

Meteorological forcing

Modified meteorological forcing

Bias correction

Regional climate modelRegional climate model

Hydrological model(independent from RCM)

Hydrological model(independent from RCM)

River runoff 1BDouble ensemble approach

-60%

-40%

-20%

0%

20%

40%

60%

80%

-25% 0% 25% 50% 75% 100% 125%

Incr

ease

of m

axim

um

mo

nth

ly Q

, %

Increase of mean annual Q, %

Impact assessment by RCM ensemble (Bērze)

Uncertainty prevails

-25%

-20%

-15%

-10%

-5%

0%

5%

10%

15%

20%

25%

-25% -20% -15% -10% -5% 0% 5% 10% 15%

Incr

ease

of m

axim

um

mo

nth

ly Q

, %

Increase of mean annual Q, %

Impact assessment by hydrological model ensemble

Uncertainty remains but is decreased

Decrease of both annual run-off and its maximum monthly value expected

0

2

4

6

8

10

12

1 3 5 7 9 11 13

Month

Mea

n m

on

thly

dis

char

ge,

m3/

s

BASIN-CTL

SHE-CTL

FIBASIN-CTL

BASIN-A2

SHE-A2

FIBASIN-A2

Seasonal analysis by hydrological model ensemble

Spring snow-melt flood significantly decreases

Autumn rainfall period extends into winter. Winter low flow disappears

Summer low flow period longer and better pronounced

Regional analysis: data series for RBD (MIKE Basin)

0

10

20

30

40

1 3 5 7 9 11 13

Me

an

mo

nth

ly r

un

off

, m

m

Month

Abava-CTL

Abava-A2

0

10

20

30

40

1 3 5 7 9 11 13

Me

an

mo

nth

ly r

un

off

, m

m

Month

Bērze-CTL

Bērze-A2

0

10

20

30

40

1 3 5 7 9 11 13

Me

an

mo

nth

ly r

un

off

, m

m

Month

Dubna-CTL

Dubna-A2

0

10

20

30

40

1 3 5 7 9 11 13

Me

an

mo

nth

ly r

un

off

, m

m

Month

Salaca-CTL

Salaca-A2

Regional analysis (MIKE BASIN)

15

25

35

45

140 180 220 260 300

Mean annual runoff, mm

Ma

xim

um

mo

nth

ly r

un

off

, m

m

Abava-CTL

Abava-A2

Bērze-CTL

Bērze-A2

Salaca-CTL

Salaca-A2

Dubna-CTL

Dubna-A2

W and N regions become similar to each other and contemporary E region

C and E regions get closer

Yearly average nutrient loads are assumed to remain the same in the A2 scenario, while their seasonal distribution have been changed.Loads of Norg, N-NH4, N-NO3, Porg, P-PO4 with monthly time-step are used as the input for the nutrient model of the Gulf of Riga.

River run-off was calculated for control period and A2 scenario by MIKE BASIN hydrological model with daily time-step. Model was set-up for the drainage basin of the Gulf of Riga, dividing it into 42 subbasins.

0

5000

10000

15000

20000

25000

Jan Feb Mar Apr Mai Jun Jul Aug Sep Oct Nov Dec Jan

Nit

rog

en

loa

d, t

on

ne

/mo

nth

CTL

A2

Nutrient run-off 1B*

ORIGINAL PLAN – 3D climatic modeling failed

Gulf of Riga: vertical temperature distribution

General Ocean Turbulence Model (GOTM) Coefficients of second order model: Cheng (2002)

Dynamic equation (k-ε style) for TKE

Dynamic dissipation rate equation

Sea state modeling 1C

Model forcing

Ins titute Model Driving data Ac ronym E xperiment

S MHI R C AO high res . HadAM3H A2 HC C T L _22 control

S MHI R C AO high res . HadAM3H A2 HC A2_22 scenario

Ins titute Model Driving data Ac ronym E xperiment

S MHI R C AO high res . HadAM3H A2 HC C T L _22 control

S MHI R C AO high res . HadAM3H A2 HC A2_22 scenario

Climate data from PRUDENCE. Control: 1961-1990, Scenario A2: 2070-2100

Extra downscaling of RCM data (bias correction via histogram equalisation): relative humidity (used variable td2m) air temperature (used variable t2m)

Original RCM data:sea level pressure (used variable MSLP)cloudiness (used variable clcov)wind speed (used variable w10m)wind direction (used variable w10dir)

Calculations made for Gulf of Riga (50 m), 30 year period, daily output data – water temperature

Physical model results – I(mean temperature distribution over depth)

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

0

0 1 2 3 4 5 6 7 8 9 10 11 12

Temperature, degC

Dep

th, m

Water 1961-1990

Water 2071-2100

Air 1961-1990

Air 2071-2100

T increase by 1,5 (bottom) to 3 (surface) degrees

Surface T increase close to air T increase

Physical model results – II(mean daily pycnocline depth and its variation)

-60

-50

-40

-30

-20

-10

0

Jan Feb Mar Apr Mai Jūn Jūl Aug Sep Okt Nov Dec

De

pth

, m

1961-1990

2071-2100

Pycnocline develops earlier...

... stratification lasts longer

Physical model results – III(mean time-depth plots of temperature)

Contemporary climate

Climate change scanario A2