revision of the svat model terra in the cosmo model based on the validation of terra standalone

Deutscher Wetterdienst

Revision of the SVAT model TERRA in the COSMO model based on the validation of TERRA standalone

J. Helmert and G. Vogel



Adaption of improvements in TERRA standalone (F. Ament - ZMAW University Hamburg) into COSMO code

Incorporation of enhanced external parameters (min. stom. resistance, emissivity, NDVI climatology, clim. soil temperature)

Adaptions of look-up tables (ensure backward compatibility)

Impact of adaptions in TERRA on soil moisture analysis (SMA) in COSMO-EU (DWD) has to be considered

Background


Simulations with TERRA standalone


CEOP sites

Lindenberg/DTongyu/China

snow, fully humid, hot summer (Dfa)

warm temperate, fully humid, warm summer (Cfb)

snow, winter dry, hot summer (Dwa)

812 mm/a

Bondville/USA

2003 2003 2003375 mm/a 312 mm/a


List of grassland/cropland sites with complete data sets

network location latitude longitude altitude climate classification vegetation availability

CEOP Lindenberg/D 52.27N 14.12E 73m asl warm temp, fully hmd, warm su grassland 2003-2006

CEOP Bondville/USA 40.00N 88.30W 216m asl snow, fully humid, hot summer cropland 2003,2004

CEOP Tongyu/China 44.40N 122.90E 184m asl snow, winter dry, hot summer grassland 2003,2004

CEOP Tongyu/China 44.40N 122.90E 184m asl snow, winter dry, hot summer cropland 2004

FLUXNET Mitra IV (P) 38.48N 8.02W 190m asl warm temp, hot and dry summer grassland 2005

FLUXNET Amplero (I) 41.90N 13.61E 884m asl warm temp, hot and dry summer grassland 2003

FLUXNET Tadham Moor (UK) 51.21N 2.83W 3m asl warm temp, fully humid, warm su grassland 2001

FLUXNET Oensingen (CH) 47.29N 7.73E 450 asl warm temp, fully humid, warm su grassland 2002-2006

FLUXNET Lille Valby (DK) 55.68N 12.08E 10m asl warm temp, fully humid, warm su grassland 2004,2005

FLUXNET Malga Arpaco (I) 46.12N 11.70E 1730m asl warm temp, fully humid, warm su grassland 2003

FLUXNET East Saltoun (UK) 55.90N 2.50W 97m asl warm temp, fully humid, warm su cropland 2004,2005

FLUXNET Lonzee (B) 50.55N 4.74E 165m asl warm temp, fully humid, warm su cropland 2005

FLUXNET Mehrstedt (D) 51.58N 10.66E 286m asl warm temp, fully humid, warm su cropland 2004-2005

FLUXNET Grignon (F) 48.84N 1.95E 125m asl warm temp, fully humid, warm su cropland 2004-2005


Tested options and assumptions

Reference conditions for different soil types (sand, silty sand, loamy sand, sandy loam and loam)

Impermeable lower hydrological boundary (“rigid lid”)

Ground water as lower boundary condition Depletion of root density with depth Bare soil evaporation according to Noilhan &

Planton (1989) Reduction of root depth (0.2m) Revised parameterization of infiltration allowing

higher infiltration rates Moisture drainage and diffusion

parameterization according to Brooks & Coorey (DWD soil types)

Soil heat conductivity does not depend on soil moisture

Satellite derived LAI and plant cover (climatolo- gical annual cycle)

Variation of surface roughness Variation of stomatal resistance and wilting

point

… and various combinations of them


CEOP Lindenberg (Falkenberg) 2003

reference run (v4.0)

measurement

experimental run

„Rigid lid“

Soil moisture

(27- 81cm)

ke_soil_hyke_soil_hy ke_soil_hy

Routine RL GW

PV




measurement

experimental run

„Root depletion with depth“

(27- 81cm)

Soil moisture




measurement

experimental run

„Bare soil evaporation according to Noilhan & Planton (1989)“

(3 - 9cm)

Soil moisture



Similar improvement also achieved in 2004 – 2006 !

Bowen ratio = Sensible heat fluxLatent heat flux


Bondville (USA) 2003 Tongyu (China) 2003

Cropland site (corn) Grassland site


Soil moisture (3 – 9cm)

Constant groundwater (lowbound=3)

vs.

Rigid lid lowbound=2 (reference)

moreover

Bare soil evaporation (Noilhan & Planton)

Root depeletion with deth (rootdep=0,70m)

satbased LAI +plcov

Impact of ground water compared to rigid lid

Falkenberg

2003 2004 2005

2006 2007 2008


Soil moisture (9 – 27cm)Constant groundwater (lowbound=3)

vs.


moreover



satbased LAI +plcov


Falkenberg

2003 2004 2005

2006 2007 2008


Soil moisture (27 – 81cm)Constant groundwater (lowbound=3)

vs.


moreover



satbased LAI +plcov


Falkenberg

2003 2004 2005

2006 2007 2008


Constant groundwater (lowbound=3)

vs.


moreover



satbased LAI +plcov

Impact of ground water compared to rigid lidBowen ratio (09-16UTC)

Falkenberg

2003 2004 2005

2006 2007 2008


COSMO-EU Experiments


Start 2009030100: #6795 (w/o SMA,HeatCond); #7004 (SMA, HeatCond)




COSMO-EU Experiments – Properties and Goals

Bare soil evaporation after Noilhan and Platon, 1989 (BS) - former Dickinson, 1984

Non-uniform root distribution (RD)

Ground water with upward diffusion

HeatCond: Soil moisture dependent heat conductivity (#7004)


Soil moisture [mmH2O]Layer 3cm-9cm

Difference: Reference – Experiment [mmH2O]

Reference

EXP 6795 Soil moisture 2009031700


Soil moisture [mmH2O]Layer 27cm-81cm

Difference: Reference – Experiment [mmH2O]

Reference

EXP 6795Soil moisture 2009031700


EXP 6795

2m-Temperature 2009031700

2m-Temperature [degC] Difference: Reference – Experiment [K]

Reference


Verification


Bias 00 UTC Runs

#6795


Quality 00 UTC Runs

#6795


Bias 12 UTC Runs

#6795


Quality 12 UTC Runs

#6795


Bias 00 UTC Runs

#7004


Quality 00 UTC Runs

#7004


Bias 12 UTC Runs

#7004


Quality 12 UTC Runs

#7004


Experiments at MeteoSchweiz

Thanks to J.-M. Bettems and P. Kaufmann


47c Test suite (COSMO modification 47c) with improved soil model

opr Operational analysis (COSMO modification 858 - 866), reference

-7

COSMO suite coarse resolution with 6.6 km grid spacing, 393 x 338 grid points

-2

COSMO suite fine resolution with 2.2 km grid spacing, 520 x 350 grid points

Verification Domains

Domain Single Versions Comparisons

47c-7opr-7 47c-2 opr-2 -7 -2

eu 47c-7@eu opr-7@eu

opr/47c-7

alps 47c-2@alps opr-2@alpsopr/47c-2

ch 47c-7@ch opr-7@ch 47c-2@ch opr-2@ch opr/47c-7 opr/47c-2

Experiments


Timeseries of T_2M and TD_2M


Comparison of diurnal cycleT_2M


Comparison of diurnal cycleTD_2M


Parameter Score 47c

7 2

eu ch alps ch

PS ME ~ ~ ~ +

STD ~ ~ ~ ~

T_2M ME - + + +

STD ~ - - -

TD_2M ME - - - -

STD ~ - ~ -

FF_10M ME - + - ~

STD ~ ~ ~ ~

CLCT ME ~ + ~ -

STD ~ ~ ~ ~

FBI 30% ~ ~ ~ ~

FBI 80% ~ + ~ ~

TOT_PREC ME ~ + ~ ~

STD ~ ~ ~ ~

+ better~ neutral- worse

PS, PSML: The test versions 47c shows a similar pressure bias as the reference version, with only few stations within the alps showing a slightly larger positive bias. The standard deviation is also almost identical. For the reduced pressure, the results are even closer between the two versions.

T_2M: The temperature is generally higher in 47c, leading to a more positive bias at the vast majority of the stations, and to less negative bias only at stations north and south of the Alps. The negative bias within the Alps remains or is even more pronounced at particular stations. The standard deviation is very similar for test and reference.

TD_2M: The dewpoint temperature bias is switching from an overestimation (wet bias) to an underestimation (dry bias) at most inland stations.

The impact on the standard deviation is neutral with only a moderate increase at few stations in the Alps.

FF_10M: The wind speed is generally increased, increasing the positive wind speed bias at many sites but decreasing the negative bias in the mountains. The impact on the standard deviation is neutral.

CLCT: The cloud coverage bias is similar in the test version 47c as in the reference. The impact on the standard deviation is small.

TOT_PREC: The positive precipitation bias of COSMO-7 is slighly reduced at few stations in the Alps only, but very similar everywhere else and also similar for COSMO-2. The standard

deviation and frequency bias are all similar.


Adaption of improvements in TERRA standalone (F. Ament - ZMAW University Hamburg) into COSMO code

Comparison and verification of experiments at DWD and MeteoSchweiz




Consideration of T_2M and TD_2M Bias – wrong effect from right reasons?

Summary & Outlook

revision of the svat model terra in the cosmo model based on the validation of terra standalone

Documents