Vers l’assimilation des données de radiomètre micro-onde sol pour la prévision du brouillard : application aux campagnes Bure 2015 et 2016

Pauline Martinet, Frédéric Burnet, Domenico Cimini, Benjamin Ménétrier, Yann Michel, Grégoire CayezAnd many others (for technical deployment)

AMA2018, 12-15 février 2018

The Bure 2016 field experiment : fog forecast improvement

■ Fog forecasting remains difficult due to uncertainties in :

- modelling of small scale processes (turbulence, microphysics)

- larger errors in the model initial state due to the undersampling of the boundary layer by current operational networks

■ An advanced use of ground-based microwave radiometers could be beneficial as they provide : continuous measurements of temperature/humidity profiles and liquid water path.

What can we expect from

the assimilation of MWR observations to

improve fog forecasting ?

From brightness temperatures (BT) to atmospheric profiles: 1D-Var retrievals

Radiative Transfer (F)ARTS/RTTOV

Model Space: T, Q, LWC

xb= Background

1h AROME forecast

BT simulation (F(x))

Observation : y

Minimization of the cost function J:

Iterative minimizationAdjustment of the background

profile

Can be run everywhere, evaluation of NWP analysis improvement

Convergence, slow (with ARTS), specification B matrix, R matrix

+-* Eriksson, P., Jimnez, C. and Buehler, S. A. 2005. Qpack, a general tool for instrument simulation and retrieval work. J. Quan. Spectrosc. Radiat. Transf. 91(1), 47-64.

F. De Angelis , D. Cimini, J. Hocking, P. Martinet, S. Kneifel: RTTOV-gb - Adapting the fast radiative transfer model RTTOV for the assimilation of ground-based microwave radiometer observations, GMD, doi:10.5194/gmd-9-2721-2016

Overview of the dataset

120 m

■ First IOP : 20161027 to 20161029

Thick radiative fog the 27th

Thin (< 50m) radiative fog the 28th

Stratus lowering not reaching the ground the 29th

■ Second IOP : 20161124-20161128

Stratus cloud lowering down to 120m the 27th

■ Third IOP : 20161201-20161203

Stratus lower during the night between 1st and 2nd reaching the ground into fog

Ceilometer cloud base

Total= 21 RS, (6 during fog), 41 UAV flights

FogStratus lowering 120m

10-27 10-28 10-29

11-24 11-25 11-26 11-27

11-02

~50/120m < 50m

120m

Data processing

■ Bias correction based on O-B monitoring

■ Background error covariance matrix sensitivity : based on an ensemble of AROME assimilation (25 members, 3.8 km) : B. Ménétrier, Y. Michel

Impact in the model and comparison with regressions : Temperature and humidity profiles

■ Large improvement of the AROME background below 700m and mainly below 250m with a decrease of the RMSE from 2.1 K to 0.5 K at 80 m

■ Almost no impact on the specific humidity humidity profile

ARO1DVAR

Bias/RMSE with respect to radiosondes

Temperature Humidite

Impact in the model and comparison with regressions : Integrated water vapor

■ Improvement of the IWV RMSE : from 1.022 kg/m² to 0.59 kg/m2

■ Improvement in the IWV temporal evolution

ARO1DVAR

Time serie of IWV difference with respect to radiosonde

Part II : Case study

Very thin radiative fog : 28102016

■ Wrong fog occurrence in the model between 1 and 5 UTC, + dissipation of the stratus cloud between 15 and 18 UTC + fog vertical extent too large

Ceilometer cloud base

Brouillard de 5 à 9 UTC

Liquid water content AROME

Stratus

Very thin radiative fog : 28102016

■ Wrong fog occurrence in the model between 1 and 5 UTC, + dissipation of the stratus cloud between 15 and 18 UTC + fog vertical extent too large

Ceilometer cloud base

Brouillard de 5 à 9 UTC

Liquid water content AROME

Stratus

1DVAR

LWP MWR = referenceAROME1DVAR

AROTower1DVAR

Temperature evolution

5 K increment

5 K increment

Intercomparison UAV/RS/MWR/AROME

RSUAV

Tower/ in-situ1DVAR ARO

28/10 08:10Thin Fog

29/10 09:48Unstable BL

Stratus 120m to 500m

28/10 15:38Unstable BLStratus cloud

Stratus lowering : night between 20161201 and 20161202

Ceilometer cloud base

Liquid water content AROME

20161201 20161202

Stratus lowering : night between 20161201 and 20161202

AROTower1DVAR

Temperature evolution

LWP MWR AROME1DVAR

20161201

5 K increment

3 K increment

2 K increment

1DVAR

20161201

Benefit of MWR to better describe the temperature profile diurnal evolution : stratus case on 20161126

1DVAR UAV

Tower/ in-situARO

26/11 07:38 26/11 08:05 26/11 08:56

Benefit of MWR to better describe the temperature profile diurnal evolution

1DVAR UAV

Tower/ in-situARO

26/11 09:47 26/11 10:16 27/11 10:48

■ Large benefit of MWR to correct the temporal evolution of the temperature profile in the first 1 km

■ Good agreement with UAV and in-situ

Conclusions

■ Improvement of the humidity retrievals with adapted vertical and cross-correlations

■ Synergy with the cloud radar BASTA to retrieve the liquid water content profile

Future Prospects

■ Significant improvement of AROME forecasts through 1D assimilation of MWR data in the lowest 500: from 2.1 K to 0.5 K RMSE

■ Temperature diurnal evolution at 50 and 120 m significantly improved after the 1DVAR (5K increment on 20161028)

■ Large improvement in the LWP initial state

■ Evolution of the vertical structure of temperature profiles improved (comparison with UAV observations)

■ Significant benefit can be expected in fog forecasting by the assimilation of MWR brightness temperatures

Main Results

Prospective campagne brouillard 2019

■ Data assimilation (1D-Var + 3D-Var or direct 3D-Var)

■ Use of the new ensemble assimilation system (flow dependent B matrix, adapted vertical and horizontal resolutions, possibility of including the hydrometeors in the control variable, extension to 4DVAR)

Future Prospects

Toulouse

Bordeaux

Super-siteLarger domain with MWR network (CNRM, ONERA, LA, Koeln, ENEA, MeteoSwiss)

Annual days of fog

Thanks for your attention