Summer School 1 - Numeric Weather

Topic 2 – Numeric Weather Prediction

NWP modelling course in the frame of Serbia for Excell Summer School was held from

June 27th to July 1st 2016 in Novi Sad, Faculty of Agriculture.

Intention of NWP modelling course was to train students to install, run and present results

of Weather Research and Forecast numerical weather prediction model. We choose WRF

as a state-of-the-art atmospheric modelling system designed for both meteorological

research and numerical weather prediction. WRF has a large worldwide community of

registered users (over 30,000 in over 150 countries).

Some of the students didn't have any knowledge and experience in numerical modeling.

It was necessary to introduce the topics like software packages, libraries and Linux basics

that are needed for the installation, running and presentation of the model products.

Comprehensive step-by-step documentation with description of necessary libraries and

installation procedure was prepared in advance and distributed to the students. Students

had individual questions that we were solving during training course. They discussed

practical issues about how to run the model, where to download the data, how to read

and plot model products.

Lateral boundary conditions for actual dates where provided and students had an

opportunity to run the model and interpret their own forecast.

Final student's discussion and presentation convinced us that all of them finished

installation and running of the model and that they understood and learned NWP

fundamentals. After this summer school all students are capable to install and run WRF on

their own. This can be a starting point for their further research in providing

meteorological forecast for crop modelling purposes.

Parameterisation of processes in soil-vegetation-atmosphere transfer schemes: Water and energy balance

H2020-TWINN-2015

a Faculty of Agriculture, University of Novi Sad, Serbia

Branislava Lalić AgMnet INTERNATIONAL

SUMMER SCHOOL IN AGROMETEOROLOGY AND CROP MODELLING

27 June – 01 July 2016

Novi Sad, SERBIA

FOCUS LECTURES II Summer School, Novi Sad, June 2016

“The understanding of turbulent transfer within foliage canopies

provides the intellectual underpinning for the physical aspects of agricultural

meteorology ...”, Kaimal & Finnigan, 1994, Atmosperic Boudary Layer flows

Penman,H.L.andLong,I.F.(1960),Weatherinwheat:Anessayinmicro-meteorology.Q.J.R.Meteorol.Soc.,86:16–50.

PLANT:FROMOBJECTTOSUBJECTincropmodelsandSVATschemes

SVAT – Soil – Vegetation - Atmosphere Transfer Schemes


MeteorologistsneedinformaTonontheheatandmoistureinputtotheatmospherefromtheearth'ssurface(e.g.,soil,vegetaTon)andhowthesurfaceextractsmomentumandkineTcenergyoftheatmospherethrough"fricTon".

BiophysicistsandecologistsneedinformaTonaboutthetemperature,

humiditylevels,solarradiaTon,cloudcover,windspeed,precipitaTon,etc.to

determinehowplantsandplantcommuniTesrespondtoenvironmental

condiTons


SVAT – Soil – Vegetation - Atmosphere Transfer Schemes Usedtomoreaccuratelydescribehowsoil,vegetaTon,andwatersurfacesexchangefluxeswiththeatmosphere MeteorologistsandclimatologistsneedinformaTonontheheatandmoistureinputtotheatmospherefromtheearth'ssurface(e.g.,soil,vegetaTon,waterbodies)andhowthesurfaceextractsmomentumandkineTcenergyoftheatmospherethrough"fricTon".BiophysicistsandecologistsneedinformaTonaboutthetemperature,humiditylevels,solarradiaTon,cloudcover,windspeed,precipitaTon,etc.todeterminehowplantsandplantcommuniTesrespondtoenvironmentalcondiTons

q Soil–threelayermodel

SVAT:SoilandVegeta9onModel


q VegetaTon-singlelayermodelGoverningequaTons

LAPS:Radia9onModule

tTCHER c

cccnc ∂∂

++λ=

ThenetradiaTonabsorbedbythe

canopy

soil

parTTonedintosensibleheat,latentheat,andstorageterms

tT

CHER ggggng ∂

∂++λ=

( ) )365C/(EHR2tT

gggngd πλ−−=

∂∂

Deepsoiltemperature

Mihailovicetal.(1992,1993,1995),J.App.Meteor

SensibleandlatentheatfluxparTToning

LAPS:Radia9onModule

( )b

acpc r

TT2cH −ρ= ( ) ⎟⎟

⎠

⎞⎜⎜⎝

⎛

+

−+

γ

ρ−=λ

cb

c

b

cpac*c rr

W1rWc

e)T(eE

( )d

agpg r

TT2cH

−ρ=

( )γ

ρ

+

−α=λ p

db

ag*sg

crre)T(e

E

( )a

rapgc r

TTcHHH −ρ=+= ( )

a

rapgc r

eecEEE −

γ

ρ=λ+λ=λ

q Sensible(Hc)andlatent(λEc)heatfluxfromvegetaTontocanopyairspace

,

q Sensible(Hg)andlatent(λEg)heatfluxfromgroundtocanopyairspace

q Sensible(H)andlatent(λE)heatfluxfromcanopyairspacetoreferencelevel

,

Governing equations

LAPS:HydrologicalModule

Waterstoredonthecanopy

Thevolumetricsoilwatercontent

ρ−=∂∂

wwfff /EPt

w

⎭⎬⎫

⎩⎨⎧

−−ρ

+−−=

∂∂

RREE

FPD1

tw

10w

l,tfg2,1l

l

1

{ }R/EFFD1

tw

2w2,tf3,22,12

2 −ρ−−=∂∂

{ }RFFD1

tw

333,23

3 −−=∂∂

“The Hitchhiker's Guide to the Galaxy” LAI ⇄ ALBEDO

Moore,K.E.,D.R.Fitzjarrald,R.K.Sakai,M.L.Goulden,J.W.Munger,S.C.Wofsy,1996:SeasonalVariaToninRadiaTveandTurbulentExchangeataDeciduousForestinCentralMassachusees,JournalofAppliedMeteorology,35(1),122-134.


Lalic, B., Fitzjarrald, D.R., Firanj Sremac, A., 2016: “Climatology” of Radiative and Turbulent Exchange at a Deciduous Forest in Central Massachusetts (manuscript in preparation)


PAR albedo 1991-2009 Solar albedo 1991-2009

Energy balance – IMPACT OF PLANTS ON ENERGY BALANCE

yxy


Energy balance – IMPACT OF PLANTS ON ENERGY BALANCE


Energy balance – IMPACT OF “ENERGY BALANCE” ON PLANTS

PredicTonsofcanopy-levelCO2assimilaTonrate(Ac)asafuncTonofLAI(panelA)andGPPasafuncTonofLAD(panelB).AcwaspredictedusingthebiochemicalmodelsofFarquharandvonCaemmerer(1982)coupledtothreedifferenttypesofcanopyfluxmodels;abig-leafmodel,amulTplelayermodel,andasun/shadebig-leafmodel.

“The Hitchhiker's Guide to the Galaxy” LAI ⇄ PHOTOSYNTHESYS CALCULATION

Monson,R.,Baldocchi,D.,2014:TerrestrialBiosphere-AtmosphereFluxes.CambridgeUniversityPress,pp.518.Firanj,A.,Lalic,B.,Podrascanin,Z.,(2014)TheImpactofForestArchitectureParameterizaTononGPPsimulaTons.Theore@calandAppliedClimatology,121,3,529-544

B


Water balance – IMPACT OF PLANTS ON WATER BALANCE

Vegetation can intercept up to 50% of the rain that falls on its leaves. The leaves of deciduous trees commonly intercept anywhere from 20 to 30% of the falling rain.

Modification of falling precipitation by vegetation The relative quantity of precipitation entering the soil is indicated in dark brown



Water balance – IMPACT OF “WATER BALANCE” ON PLANTS

EvapotranspiraToncoefficient


Water balance – IMPACT OF “WATER BALANCE” ON PLANTS


Air flow - IMPACT OF AIR FLOWS ON PLANTS

TURBULENT DIFUSION ⇄ Venturia inaequalis spatial distribution

Summer school lessons about

building SVAT models

H2020-TWINN-2015

Faculty of Agriculture, University of Novi Sad, Serbia

Dragutin T. Mihailović

AgMnet INTERNATIONAL SUMMER SCHOOL

IN AGROMETEOROLOGY AND CROP MODELLING

27 June – 01 July 2016

Novi Sad, SERBIA

Summer school lessons about building

SVAT models Dragutin T. Mihailović

Lesson 1 Epistemological features

CROP MODELLING Summer School, Novi Sad, June 2016


Problems in building SVAT models a)  How we see the object of interest? b)  How to make proper approximations? c)  Mathematical problems d)  Scalling problems


Problems in building SVAT models a)  How we see the object of interest? b)  ...................................................... c)  ....................................................... d)  .......................................................


People “naively” say: Model is the only way to approach to reality! NO! Formalization of theory cannot be older than the theory itself!!

Model is the only way to simulate a reality! Icarus


Mihailović, D.T., I. Balaž and D. Kapor, 2016: Time and methods in modelling environmental interfaces: personal insights. Elsevier, Amsterdam (in publishig dept.)

Mihailović et al., 2006, J. App. Meteorol.


What we see as a reality?

Is it a: 1) duck 2) rabbit

or 3) both


MinimumofprocessesEvapotranspira2onPrecipita2onCanopyintercep2onThroughflowStemflowIntercep2on&inflitra2on

Our reality


Problems in building SVAT models a)  .......................................................... b)  How to make proper approximations? c)  .......................................................... d)  ..........................................................


The most famous approximation – material point


bird's-eye view seen by bugs CROP MODELLING

Summer School, Novi Sad, June 2016

HowPicasso“saw”awomen1901-1968?

Howwecan“see”atreeinSVATmodels?


Leaf

Steam

Root

Portential / resistance Evapotranspiration =

Analogy to electornic circuit


Canopy

Air

Air

H

h


Why does a train clatter when it’s

moving along the tracks?




Problems in building SVAT models a)  ........................................................... b)  ........................................................... c)  Mathematical problems d)  ...........................................................


Mihailović and Eitzinger, 2007, Ecol. Modelling



Chaotic behaviour of model surface variables

Mimić and Mihailović, 2013, Modern Phys. Lett.


Problems in building SVAT models a)  .......................................................... b)  .......................................................... c)  .......................................................... d)  Scaling problems


Scalling problem is one of the major concern in building SVAT models

ItisthelackofinformaBononhowonecanaggregateand/ordisaggregatevarioussurfacefluxes(heat,momentumandwatervapor)fromonescaletoanother.


GCMmodelresolu2on

Topography

Vegeta2on

Soil

Aggrega2

on

Disaggrega2o

n

RgCMmodels

OceanLand

GCMisrunonthecoarsegridwhileLSSonthefinersubgridinputparameterstoLSSneedstobedisaggregated


Disggregation of fluxes

Mihailović et al., 2016, Science Total Enviro.


Aggregation of fluxes (Schmidt’s paradox)

The grid-box over the Prospect park NY, USA illustrating the heterogeneous grid-box used in environmental model simulations.

Mihailović et al., 2015, App. Math. Comp.

Schematic diagram illustrating how the sub-grid scale surface patch-use classes are treated within the different parameterization scheme


Schmidt’s paradoxLowestmodellevel

Averaged sensible heat flux

Single sensible heat fluxes

Countergradient



H2020-TWINN-2015





27 June – 01 July 2016

Novi Sad, SERBIA



Lesson 2

An example with the LAPS scheme



Content a)  Introduction b)  The main features of the LAPS scheme c)  Comment of outputs d)  Uncertainities in modelling procedure


Content a)  Introduction b)  ............................................................ c)  ............................................................ d)  ............................................................


How LSS works?

Soil

Reference level


Content a)  ............................................................... b)  The main features of the LAPS scheme c)  ............................................................... d)  ...............................................................

L A P S

LAPS: Soil-Vegetation-Atmosphere-Transfer model Forcing data q  Air temperature

q  Short wave radiation

q  Long wave radiation

q  Humidity

q  Wind speed

q  Precipitation

q  Initial conditions

q  Aerodynamic ch.

q  Morphological ch.

Prognostic variables Temperatures:

leaf, ground and deep soil temperature

q  Soil moisture content at three soil layers

q  Leaf wetness Main outputs Sensible and latent heat fluxes vegetation - canopy air space

ground - canopy air space canopy air space - reference level

q  Momentum flux q  Wind within the canopy

Serbian-French Project Meeting, 23 June 2008, Avignon (France)

LAPS: resistance representation

q  Mihailovic et al. (1992), J. App. Meteor. q  Mihailovic et al. (1993), J. App. Meteor. q  Mihailovic (1994), Workshop, McQuary Univ., Sydney q  Mihailovic et al. (1995), J. App. Meteor. q  Mihailovic (1996), Global and Planetary Chang. q  Mihailovic and Kallos (1997), Bound. Layer Met. q  Mihailovic et al. (1999), Bound. Layer Met. q  Mihailovic et al., (2006), J. App. Meteor.


q  Soil – three layer model

LAPS: submodels used

q  Vegetation-single layer model


Governing equations in the LAPSThe net radiation absorbed by the canopy and soilis assumed to be partitioned into sensible heat,latent heat, and storage terms, as

tT

CHER ffffnf ∂

∂++λ= (1) Prognostic

tT

CHER ggggng ∂

∂++λ= (2) Prognostic

Deep soil temperature

( ) )365/(2 πλ−−=∂∂

CEHRtT

gggnetg

d

(3) Prognostic


AGRIDEMA, Vienna, 21 November – 2 December 2005 Page 1, DTM

Governing equations in the LAPS

Water stored on the canopy

ρ−=∂

∂wwff

fEPt

w/ Prognostic

The volumetric soil water content

,11,2 0 1

1

1 g tf ll

w

E Ew P F R Rt D⎧ ⎫+∂

= − − − −⎨ ⎬∂ ρ⎩ ⎭

{ }REFFDt

wwtf 22,3,22,1

2

2 /1−ρ−−=

∂

∂

{ }RFFDt

w333,2

3

3 1−−=

∂

∂ Prognostic



Content a)  ............................................................... b)  ............................................................... c)  Comment of outputs d)  ...............................................................

Outputs

Ground surface temperature Tg Leaf temperature Tl Water intercepted by leafes wl Soil moisture content in three layers wl, w2 and w3

15

20

25

30

35

Folia

ge te

mpe

ratu

re, T

f (o C

)

Sunflower (NS-Velja)Rimski Sancevi (Serbia)

Simulated (solid line)Observed (square)

(b)

0000 0400 0800 1200 1600 2000 2400 Hours (LST)

Daily variations (19 July 1998) of the sunflower foliage temperature: (1) simulated by the LAPS model and (2) measured (Rimski Sancevi, SERBIA) (Mihailovic and Eitzinger, Ecol. Modell., 2007)

NCEP LAPS


150 151 152 153 154 155Day of Year

-100

0

100

200

300

400

500

Late

nt h

eat f

lux

(Wm

-2)

OLDNEW

Observed

(a)

Temporal variation of latent heat flux andobtained by the LAPS scheme comparedwith the observations over a soybean field inCaumont site (France) during its growingseason in 1986(Mihailovic et al., 2006, J. App. Meteor.)

0 2 4 6 8 10 12 14 16 18 20 22 24Hours (UTC)

-100

-50

0

50

100

150

200

250

300

Surf

ace

fluxe

s (W

m-2

)

Apple orchard16 May 2002Chlewiska (Poland)

SIMULATED

Sensible

OBSERVED

(a)

Sensible

Latent Latent

Diurnal variations of latent andsensible heat fluxes over(a) apple orchard, (b) rapeseed in Chlewiska (Poland) for 16 May 2002(Mihailovic et al.,Ther. App. Climat.– in revision)Measured values provided by thePoznan Agric. University Group

0 2 4 6 8 10 12 14 16 18 20 22 24Hours (UTC)

-100

-50

0

50

100

150

200

250

300

Surf

ace

fluxe

s (W

m-2

)

Rapeseed16 May 2002Chlewiska (Poland)

SIMULATED OBSERVED

Sensible

Latent

(b)

Sensible

Latent


188 190 192 194 196 198 200Julian day

0

10

20

30

40

Air t

empe

ratu

re in

can

opy,

Ta

(o C)

Simulated (thin)Observed (thick)

Ten-day variation (8-17 July 2002) of within-canopy air temperature simulated by the LAPS and observed inside a sunflower canopy at the Rimski Sancevi (SERBIA) (Mihailovic and Eitzinger, Ecol. Modell., 2007)

NCEP LAPS


Water transfer modelling: Soil

Daily averages of total soil water content (mm) over a depth of 1.6 m simulated by the LAPS and observed beneath a soybean canopy in HAPEX experiment at the Caumont (France) (Mihailovic et al., 1998, J. Hydrol.)

0 100 200 300Julian day

200

400

600

Observed (squares) Calculated (solid line)

θSo

il m

oist

ure,

t(m

m)

Field capacity (480 mm)

(WRF Modelling System - NMM) + LAPS (forcing data)



Content a)  ............................................................... b)  ............................................................... c)  ............................................................... d)  Uncertainities in modelling procedure

The energy balance equation is valid on LOCAL and GLOBAL scale (Rosenberg, 1983) In order to calculate surface temperature Ts we can use one of the following methods 1) solving this transcendental equation for Ts 2) solving this equation in F.D. form

Rn (Ts) + S(Ts) + LE (Ts) + PS (Ts) + M (Ts) = 0

Cg ∂Ts/ ∂t = Rn (Ts) - S(Ts) - LE (Ts) - PS (Ts) - M (Ts)

( ) ( ) ( )rgDrgHrgLrggg TTCTTCeTECTTatTC −−−−−−−=∂∂ ])([/

Net radiation

Latent heat Sensible heat

Soil heat

Transfer coefficient

radiation wat. vap. press. heat in

atmosp. soil heat

ground surface TEMPERATURE at reference level

surface capacity

))/(//(1 ng

ng

nng

ng TFCtFTT δδ+Δ+=+

in this form n time level Δt time step solution

backward scheme in time

5 10 15 20 25 30Local time (h)

15

20

25

30

Soil

surfa

ce te

mpe

ratu

re, T

g (C

)

6 July 1982NS-10

Modeled (solid)Observed (square)

(Mihailovic and Lalic, 1995)

0 20 40 60 80 100 120 140 160 180 200Time lag

0.0

0.2

0.4

0.6

0.8

1.0

X

Solution of energy balance equation plotted as the renormalised “effective”surface temperature X in terms of time.

00 /)( gnggn TTTX −= , where 0

gT is the initial“effective“ surface temperature.

300 350 400 450 500 550DOY (27 October 2005 - 26 June 2006)

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Frac

tion

al c

over

of

vege

tation

Reference

Error 2%

Error 5%

Error 10%

0.0 1.0 2.0 3.0 4.0 5.0 6.0Leaf area index (LAI)

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Frac

tion

al c

over

of

vege

tation

Reference Error 2%

Error 5%

Error 10%

Error in estimation of vegetation fractional cover

Avignon (France) Wheat (2005/2006)

Δσ =(1-exp(-k*LAI*r))/exp(k*LAI) K=0.34 r = ΔLAI/LAI


300 350 400 450 500 550DOY (27 October 2005 - 26 June 2006)

0.00

0.10

0.20

0.30

0.40

0.50H

eigh

ts (m

)

Canopy height (M)

Displacement height (C)

Roughnes length (C)

Canopy source height (C)



300 350 400 450 500 550DOY (27 October 2005 - 26 June 2006)

0.0

1.0

2.0

3.0

4.0

5.0

6.0

Evap

orat

ion

(mm

)

LAIError in estimation of VEG_COVBLACK referenceRED -2%BLUE -5%



300 350 400 450 500 550DOY (27 October 2005 - 26 June 2006)

-5

0

5

10

15

20

25

30

Leaf

tem

pera

ture

(deg

C)

LAI

300 350 400 450 500 550DOY (27 October 2005 - 26 June 2006)

-5

0

5

10

15

20

25

30

Insi

de c

anop

y te

mpe

ratu

re (d

eg C

)

LAI


( Theor.App. Clim., Mihailovic et al., 2008)

Calculation performed according to LAPS scheme



H2020-TWINN-2015





27 June – 01 July 2016

Novi Sad, SERBIA



Lesson 3

Links with other models


Regional climate 3- D simulations

Temporal ). 651 grid points

Mihailovic et al., 2004, JAM

Surface temperature

).

0 6 12 18 24Hours (GMT)

0

5

10

15

20

25

Surf

ace

tem

pera

ture

(de

g C)

(blue ) Tall grass (Tmin=12.1, Tmax=18.0)(pink ) Crop (Tmin=11.5, Tmax=21.0)(orange) Mixed woodland (Tmin=12.0, Tmax=17.0)(black ) Control (Tmin=11.8, Tmax=19.6)

Sensitivity of mean surface temperature to cover type. The hourly means were made using all grid points in the domain of interest.

Mihailovic, D. T., 2003: Implementation of Land-Air Parameterization Scheme (LAPS) in a limited area model. Final Report, The New York State Energy Conservation and Development Authority, Albany, NY, 110 pp.

Air temperature at 2m

).

0 6 12 18 24Hours (GMT)

0

5

10

15

20

25

Air

tem

pera

ture

at

2m (

deg

C)

(blue ) Tall grass (Tmin=12.3, Tmax=17.6)(pink ) Crop (Tmin=11.7, Tmax=19.8)(orange) Mixed woodland (Tmin=12.1, Tmax=16.9)(black ) Control (Tmin=11.9, Tmax=18.7)

The hourly means were made using all grid points in the domain of interest.

Mihailovic, D. T., 2003: Implementation of Land-Air Parameterization Scheme (LAPS) in a limited area model. Final Report, The New York State Energy Conservation and Development Authority, Albany, NY, 110 pp.

3) Chemical Transport Models

5th International Summer School “Renewable Energy & Energy Efficiency in South East Europe”

10 – 21 August 2009, Fojnica, Bosnia and Herzegovina

Chemical transport models While related general circulation models (GCMs) focus on simulating overall atmospheric dynamics (e.g. fluid and heat flows), a CTM instead focuses on the stocks and flows of one or more chemical species

iEMSs 2006 July 9-13, 2006 The Wyndham Hotel, Burlington, Vermont, USA Page 10

Domain of integration





Mihailović et al., Environ. Sci. Poll. Res. Int. (2008, 2009) To activate EMEP with new meteorology (advection, surface processes and boundary layer)





6) Hydrological Models



Water transfer modelling: Soil

Daily averages of total soil water content (mm) over a depth of 1.6 m simulated by the LAPS and observed beneath a soybean canopy in HAPEX experiment at the Caumont (France) (Mihailovic et al., 1998, J. Hydrol.)

0 100 200 300Julian day

200

400

600

Observed (squares) Calculated (solid line)

θSo

il m

oist

ure,

t(m

m)

Field capacity (480 mm)



7) Forecasting the Occurrence of Plant Diseases



BAHUSbiometeorological system for predicting the occurrence of diseases inplant production (Center for Meteorology and Envoronmental Predictions)

1) Clamatological data 2) Automatic weather station 3) NCEP outputs (mesoscale) 4) LAPS surface scheme

(Mihailovic et al., Envir. Modell. Soft., 2001)

Venturia inequalis Ervinia amylovora Plasmopra vitiocola

FP6 project ADAGIO


8) Modelling of the UV radiation



NEOPLANTA computes the: 1) solar direct and diffuse UV irradiances under cloud free conditions for the wavelength range 280–400 nm and UV index 2) effects of O3, SO2, NO2, aerosols, and nine different ground surface types on UV radiation are included. 3) instantaneous spectral irradiance for a given solar zenith angle 4) UV index for the whole day at half-hour intervals from sunrise to sunset Modelling details: 5) atmosphere is divided into several parallel layers (maximum 40) in the model 6) it is assumed that the layers are homogeneous with constant values of meteorological parameters 7) vertical resolution of the model is 1 km for altitudes below 25 km and 5 km above this height. 8) upper boundary of the highest layer in the model is 100 km 9) model uses standard atmosphere meteorological profiles 10) there is also an option of including the real-time meteorological data profiles from the high-level resolution mesoscale


NEOPLANTA: A Short Description of the First Serbian UV Index Model Authors: Malinovic, S., Mihailovic, D.T., Kapor, D., Mijatovic, Z., Arsenic, I.D., Publication: Journal of Applied Meteorology and Climatology, vol. 45, Issue 8, p.1171-1177



Documents

Summer School 1 - Numeric Weather