EMS2013 09 - 13 Sep Reading UK Evaluating modelled surface long wave downward radiation with Cabauw observations: The GABLS3 SCM case. Fred Bosveld (KNMI)

EMS2013 09 - 13 Sep Reading UK

Evaluating modelled surface long wave downward radiation with Cabauw observations: The GABLS3 SCM

case.

Fred Bosveld (KNMI)

Content:

• Motivation

• An improved formula for Long Wave Downward radiation

• High quality validation data

• Optimization of Improved formula

• Application in GABLS3 SCM

• Conclusions

EMS2013 09 - 13 Sep Reading UK

Motivation (1)

GABLS3 SCM case01-July-200624 hoursmoderate geowind (7 m/s)

Emphasis on stable boundary layer and the transition periods

Intercomparison and evaluation of 19 Models

EMS2013 09 - 13 Sep Reading UK

Motivation (2)

Long wave downward radiationat the surface:

Models spread and underestimate

Models differ in:

- Thermodynamic profiles- Profiles of GHG- Radiation parameterization

How to evaluate?

- Get rid of the influence of Thermodynamic profiles!

EMS2013 09 - 13 Sep Reading UK

Strategy

Start with Brunt (1932) L↓ as function of 2-m temperature (T2) and water

vapour pressure (e2)

Extend simple formula of Brunt (1932) with profile information

Optimize against high quality observations.

Half of L↓ at the surface originates from lowest 200 m (Ohmura, 2001)

Temperature gradient is important for L↓ (Dürr and Philipona, 2004)

422 )( TebaL

EMS2013 09 - 13 Sep Reading UK

Observations from CESAR:Cabauw Experimental Site for Atmospheric Research

Observations of Cabauw (NL) (Boers et al., 2011)

Clear Sky detection very strict :

Nubiscope – low and middle cloudsCloud radar – cirrus clouds

L↓ from Cabauw BSRN station:

Accuracy 2-3 W m-2

1 year of data (May 2008 – Apr 2009)

EMS2013 09 - 13 Sep Reading UK

Results (Brunt, 1932)

422 )( TebaL

Optimized parameters

a = 0.660± 0.002

b = 0.0419± 0.0007

Brunt deviates at day and nightbecause upper air is warm at

nightand cool at day time

EMS2013 09 - 13 Sep Reading UK

Results (Extensions of Brunt, 1932)

fTebaL 422 )(

Separate Day and Night

fTTedcTebaL )()()( 42

4200200

422

422 )( TebaL

Substitute part of surface radiation by radiation emitted at temperature of 200 m

fTTedcTebaLref )()()( 42

4200200

42200

Use e200 instead of e2

EMS2013 09 - 13 Sep Reading UK

Results (Optimizations)

10 minute mean clear sky observations

N = 2450 allN = 1480 nightN = 600 day

Brunt (1932) flavours

Sig (y)W m-2

Original 7.57

Add offset

All 6.32

Day 5.43

Night 5.91

Add T200

All 4.62

Day 4.38

Night 4.46

Change e2 -> e200

All 4.14

Day 3.94

Night 3.96

Add offset : improved but significant difference between night and day

Add T200 : Improved and much less difference between night and day

Change e2 -> e200 : Improved further

EMS2013 09 - 13 Sep Reading UK

Results (Improved and more general fit)

Brunt Extended deviations

almost gone because upper

air temperature is takeninto account

Brunt deviates at day and night

because upper air is warm at night

and cool at day time

EMS2013 09 - 13 Sep Reading UK

Coming back to GABLS3 SCM

Under estimation almost disappears after correcting for the effect of

thermodynamic profiles on L↓.

ΔL↓= L↓ - L↓ref

0400 UTC

EMS2013 09 - 13 Sep Reading UK

Conclusions

T

An extension of the Brunt (1932) LWD formula is found:• More general in time than the original simple formula• Residuals in the fit approaches the high accuracy of the

observation• The effect of e and T profile above 200 m is only statistically

taken into account.• Thus the coefficients are most likely location dependent.

Application in GABLS3 SCM case:• LWD bias in most Models explained• Thermal coupling to the surface dominates T-profile. (Bosveld

et al., 2013, BLM (In review)).

See Holtslag et al. (2013) (BAMS) for GABLS overviewSee www.cesar-observatory.nl for information and data of CESAR