Slide 1
Assimilation of near real time GEOV Albedo and Leaf Area Index
within the ECMWF modelling system
Souhail Boussetta, Gianpaolo Balsamo, Emanuel Dutra, Anton Beljaars
Slide 1BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
with the support and partnership of the EU FP7 project
Slide 2
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 2
Why vegetation state is important?Vegetation was shown to be of critical importance under the NWP framework:
Evapotranspiration Boundary layer developement Cloud and precipitation ...
Vegetation directly affect the global carbon cycle
Earth System Models are evolving to better represent vegetation dynamic
Satellite observations with channels informative on the vegetation state are becoming more and more available and with higher accuracy & frequency
Assimilation of vegetation state products would allow:
Better represent land biogenic fluxes in interaction with the atmosphere Monitor present and past vegetation state and its dynamics Understand and adjust process development within models Ultimately to improve weather and earth systems prediction.
Slide 3
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 3
The data:The Copernicus GEOV1 LAI/albedo product is based on observations from the VEGETATION sensor on board of SPOT satellite. Global coverage with 1km / 10 day space/time resolutionProduced in the framework of the Copernicus Initial Operation & IMAGINES and Freely available.
The model: The ECMWF LSM (CTESSEL) coupled within the Integrated Forecasting system IFS
The analysis system:The analysis procedure is an optimal combination of the satellite observations and derived climatology, depending on their associated errors σo and σc. Suitable for NWP framework and consistent with actual method used for slow evolving
variables.
clim
Analysed trajectoryobs
var
t
ccooa varvarvar
22
2
22
2
,co
oc
co
co and
where
How vegetation state is take into account?
Slide 4
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta Slide 4
Observation
Quality control + sf (albedo)
GEOV1 LAI/Alb
9-point Spatial smoothing
GEOV1 Observation + o
Analysis
ccooa CObsAna lim
22
2
22
2
,co
oc
co
co and
GEOV1 Analysis
Climatology(a-priori)
Quality control + sf (albedo)
GEOV1 LAI/Alb
Averaging => Gapped Climatology
Temporal Gap filling
Extraction of 10-day look-up table on land use type
GEOV1 Climatology + c
9-point Spatial smoothing
Spatial Gap filling
Vegetation state initialization
Slide 5Europe East China 2003droughts
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 5 NRT analysed LAI seems able to detect/monitor anomalous year The analysed LAI and albedo signal appear to be covariant mainly during wet year.
Horn of Africa 2010 drought
Australia drought recover also visible in albedo
LAI anomaly Albedo anomaly
Russian 2010 Heat wave
In which cases vegetation matters most?
Slide 6
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 6
The surface-only simulation setup:To seek the impact of the NRT analysed data four experiments are performed
Period: 1999 to 2012 Coverage: Global Resolution: 40km
4 different experiments:Control: LAI+albedo climatology are usedNRT_ALB_LAI: LAI nrt data + albedo nrtNRT_LAI: LAI nrt data + albedo climatology NRT_ALB: LAI climatology + albedo nrt
Results evaluated on surface fluxes:Latent heat flux Sensible Heat fluxCarbon dioxide flux (Net Ecosystem Exchange of COt)
Which processes are affected by vegetation state?
Slide 7
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 7
Clim
NRT_ALB - Clim NRT_LAI - Clim
NRT_ALB_LAI - Clim
Latent heat flux sensitivity
Slide 8
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 8
Clim
NRT_ALB - Clim NRT_LAI - Clim
NRT_ALB_LAI - Clim
Sensible heat flux sensitivity
Slide 9
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 9
Clim
NRT_ALB - Clim NRT_LAI - Clim
NRT_ALB_LAI - Clim
Carbon dioxide flux sensitivity
Slide 10
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 10
The atmospheric coupled simulation setup:To seek the impact of the NRT analysed data four coupled experiments are performed
Period: 2010 Coverage: Global Resolution: 40km
4 different experiments:Control: LAI+albedo climatology are usedNRT_LAI: LAI nrt data + albedo climatology NRT_ALB: LAI climatology + albedo nrt NRT_ALB_LAI: LAI nrt data + albedo nrt
Results evaluated on weather forecasts for next day :2m temperature (36-hour forecast range)2m relative humidity (36-hour forecast range)
Is weather affected by vegetation state?
Slide 11
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 11
Horn of Africa drought & Australia drought recover
LAI anomaly Albedo anomaly
Check the T 2m and RH on short term forecast fc+36 valid 12 UTC, Nov. 2010Sensitivity = (exp –ctl),
if >0 => warming/adding moisture, if <0 => cooling/removing moisture
And Impact = |ctl – analysis| - |exp – analysis| ,
if >0 => relative error reduction from the analysis (positive impact ) if <0 => relative error increase from the analysis (negative impact)
Is weather affected by vegetation state (II)?The focus is on:
Slide 12
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 12
Coupled run: NRT_LAI+ALB Vs ClimSensitivity Impact
Weather forecasts sensitivity and impact
ΔRH2m
Red Warming
Blue Drying
Blue Better forecast
Blue Better forecast
ΔT2m ΔErr T2m
ΔErr RH2m
Slide 13
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 13
Conclusions & OutlookThe proposed analysis procedure leads to smooth temporal evolution of LAI/Albedo, which is more appropriate for initialization of environmental prediction models.
The analysed NRT LAI is able to detect/monitor anomalous year. The analysed NRT albedo signal seems covariant with the NRT LAI mainly during wet year (compensation effect may occur between vegetation and bare-ground albedo).
GEOV1 NRT LAI/albedo showed potential for heat waves and drought monitoring but given the impact on energy and carbon fluxes these products are ECVs also for coupled experiments.
Introducing NRT LAI and Albedo in coupled runs is physically justified and has an overall neutral to positive impact on forecasted weather parameters, with the LAI signal being dominant and in some cases enhanced with the Albedo anomaly signal.
In future work, enhanced connections between albedo, LAI (and roughness) in Earth System Models will most likely increase the sensitivity to vegetation dynamics.
Slide 14
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 14
Thank you for your attention
Contact: [email protected]
http://fp7-imagines.eu/
Slide 15
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 15
Horn of Africa drought & Australia drought recover
LAI anomaly Albedo anomaly
Slide 16
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 16
T2m Sensitivity Impact
Rh2m
Yellow Warming Blue Positive impact
Blue Positive impactBlue drying
Coupled run: NRT_LAI Vs Clim
Slide 17
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 17
Coupled run: NRT_ALB Vs ClimT2m Sensitivity Impact
Rh2m
Slide 18
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 18
LAI anomaly Albedo anomaly
2010 Russian Heat wave
Slide 19
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 19
Latent Heat flux
Clim
NRT_ALB - Clim NRT_LAI - Clim
NRT_ALB_LAI - Clim
Slide 20
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 20
Sensible Heat flux
Clim
NRT_ALB - Clim NRT_LAI - Clim
NRT_ALB_LAI - Clim
Slide 21
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 21
Net Ecosystem Exchange
Clim
NRT_ALB - Clim NRT_LAI - Clim
NRT_ALB_LAI - Clim
Slide 22
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 22
Scores of GEOV1 LAI NRT against GEOV1 LAI climatology for JJA: a) 2m temperature sensitivity [K], b) 2m temperature impact, c) 2m relative humidity sensitivity [%], d) 2m relative humidity impact. An overall neutral to positive impact.
Assimilation of GEOV1 NRT LAI and its potential value (coupled runs)
ctlLAILAI XXXysensitivit _exp_)( ||||)( exp__ anLAIanctlLAI XXXXXimpact
T2m
RH
Blue cooling
Blue drying
Blue Positive impact
Blue Positive impact
Slide 23
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 23
2003 Europe and East China drought
LAI anomaly Albedo anomaly
Slide 24
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 24
Latent Heat flux
Clim
NRT_ALB - Clim NRT_LAI - Clim
NRT_ALB_LAI - Clim
Slide 25
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 25
Sensible Heat flux
Clim
NRT_ALB - Clim NRT_LAI - Clim
NRT_ALB_LAI - Clim
Slide 26
BG2.5 EGU, Vienna 02 May 2014. S. Boussetta
Slide 26
Net Ecosystem Exchange
Clim
NRT_ALB - Clim NRT_LAI - Clim
NRT_ALB_LAI - Clim