Land Surface Analysis SAF:

Contributions to NWP

Isabel F. Trigo

Scope of Presentation

I – LSA SAF

- Products relevant for NWP

- Schedule

- Beta – Users

II – LSA SAF / ECMWF joint Collaboration

- Assimilation of LST ?

- ECMWF Land Surface Scheme

- Future Work

I – Land Surface Analysis SAF

Satellite Application Facilities

Development & Processing Centres within the EUMETSAT Applications Ground Segment:

Meteorological products

Software Packages

Using MSG and EPS data.LSA SAF

Consortium of 13 organisations / 8 member states

Host Institute: IM

LSA SAF PRODUCTS

Priority 1 & 2 Benefits for NWP

Snow Cover Validation; Albedo of Snow Areas;

Initialisation of snow mass in DA

Albedo Broadband/Spectral albedo updates

Land Surface Temperature Validation; DA

Sfc Short-wave Fluxes

Sfc Long-wave Fluxes

Validation; forcing for global soil moisture AN.

LSA SAF PRODUCTS

Priority 3 & 4 Benefits for NWP

Soil Moisture

Evapotranspiration

Validation

Vegetation Cover

LAI

Possible use for monthly/seasonal updates

2003 2004 2005 2006 2007 2008

DevelopmentPre-operational Operational

Snow Cover

white - class “snow”;light grey -class “partially snow covered”; green class – class “clear land”;

Coverage – Europe; (Global)

Resolution – 5x5 km; (1x1km)

15min for MSG; daily maps

Delivery Time - Near Real Time

2003 2004 2005 2006 2007 2008

DevelopmentOperational

Beta-Users

Albedo

0.30

0.15

0

Coverage – MSG disk; (Global)

Resolution – 3x3 km; 10 days

Delivery Time - Near Real Time

Content – Spectral & Broadband:

VIS 0.6; VIS 0.8; IR 1.6

300-4000 nm; 400-700nm; 700-4000nm

Accuracy - ~1%

Broadband Albedo

2003 2004 2005 2006 2007 2008

DevelopmentOperational

Beta-Users

Land Surface Temperature

Coverage – MSG disk; (Global)

Resolution – 3x3 km; 15 min

Delivery Time - Near Real Time

Accuracy – ~1 – 2 K

2003 2004 2005 2006 2007 2008

DevelopmentOperational

Beta-Users

II – LSA SAF / ECMWF Collaboration(Pedro Viterbo and Anton Beljaars)

Assimilation of LST ?

Window Channel Brightness Temperatures (10.50 – 12.50 m)

METEOSAT – 7 clear sky .......... Tb_OBS

ECMWF model equivalent .......... Tb_ECMWF

To assess the quality of modelled land surface temperature.

To monitor the quality of the observations by identifying gross errors / systematic problems in the data.

Tb_obs Tb_ECMWF

averaged for 1-15 Feb 2001

Overestimation Tb diurnal cycle, mostly along the ITCZ

Underestimation of Tb daily cycle over arid/semi-arid regions

Sources of Window channel Tb Systematic Errors:

(i) cloud screening of the observations ...

- threshold for the %Clear Sky pixels;- MSG will allow more accurate cloud masks.

(ii) underestimation of modelled diurnal amplitudes of Tb_ECMWF in clear sky conditions

underestimation of LST diurnal cycle

(particularly over arid and semi-arid regions)

Further Details: Trigo and Viterbo (2003), J. Appl. Met., 42, 1463-1479.

A pre-requisite to contemplate assimilation: The model has to look like the observations

Window Channel Tb bias (and thus LST bias) ~ 10K in semi-arid regions strongly limits the assimilation of:

Window channel TOA Radiances / LST product;

TOVS channels sensitive to the lower troposphere, over land.

A set of sensitivity experiments of Tb amplitude to surface parameters / variables pointed

Model deficiencies in the

LAND – ATMOSPHERE Aerodynamic Coupling

ECMWF Land Surface Model - TESSEL

RaRa,nv

Tc

Tskin

Dry Vegetated Surfaces

gbarehighVlowV

highVhighVlowVlowV

Hff

HfHfH

_1

Bare Ground Tile:a

skinairpgbare R

TTcH

_

Ra does not take into account the wind shielding effect of surrounding vegetation.

Hbare_g is overestimated, particularly over sparsely vegetated areas.

Ra

Rs

Tskin

Ra,nv

Tc

Revised TESSEL

sa

skinairpgbare RR

TTcH

_

Rs – Resistance to turbulent heat transfer within the under canopy layer.

- sparseness of the vegetation;

- canopy height;

- ...

2*

1

ubaRs

ECMWF 1-Column Model

1 2 3 4 5 6 7 8 9 10 11 12 13 14 155

10

15

20

25

30

35

40

Tb -A

mp

(K

)1-15 FEB 2002;26N08E

ATbobs

ATbctrl

ATbnew

0 2 4 6 8 1010

15

20

25

30

35

Tb -A

mp

(K

)

Wind Speed 10m (ms-1)

1-15 FEB 2002;26N08E

OBSCTRLNEW

Western Sahara – Validation Period

Parameters in Rs

1) Estimated by fitting daily Tb_ECMWF amplitudes to Tb_OBS; 1-15 Feb 2001;

2) Fixed for each grid point

1-15 Feb 2002.

Calibration Validation

CTRL NEW CTRL NEW

RMSE 7.0 K 2.2 K 7.2 K 2.2 K

BIAS -6.3 K <0.1 K -6.5 K -0.2 K

Tb Daily Amplitude

Summary

I – LSA SAF

• Priority 1 Products: Snow Cover, Albedo, and Land Surface Temperature

• Priority 2 Products: Downwelling Surface Short & Long wave Fluxes

• Priority 3- Vegetation Parameters: FVC, LAI- Soil Moisture / Evapotranspiration

• Beta-Users - Please contact LSA SAF NOW !- Pre-Operational in 2005

• Operational Phase - 2007

II – LSA SAF / ECMWF Collaboration

• The comparison of Meteosat-7 window channel Tb (clear sky) with Tb_ECMWF shows that the modelled Tskin diurnal cycle is underestimated over arid / semi-arid regions.

Limits assimilation of LST & Channels that peak in the lower

troposphere.

• Sensitivity experiments pointed to deficiencies in the Land Surface – Atmospheric aerodynamic coupling.

TESSEL Bare Ground Tiles ignore wind shielding effect of the

surrounding vegetation.

• Introducing an extra resistance Rs to turbulent heat transfer within the canopy layer substantially reduces the ECMWF cold bias of Tb daily amplitudes.

Next

• Estimate Rs parameters for Meteosat (MSG) disk (by fitting daily amplitudes of Tb_ECMWF to the observations).

• Study the impact of the improved land Tskin in the assimilation of TOVS channels sensitive to the lower troposphere.

• Test 1D-var assimilation of window channels TOA radiances/ retrieved LST to further improve the model Tskin over land.