Title: Numerical prediction of fog for airports : local assimilation, nowcasting and predictability
1Numerical prediction of fog for airports local
assimilation, nowcasting and predictability
- Samuel Rémy, Stevie Roquelaure and Thierry Bergot
(Météo-France) - FRAM workshop on fog, 21-22 May 2008
2Outline
- Description of the model
- Operationnal use at Paris-CdG
- Current research on COBEL-ISBA
- Local Ensemble Prediction System (L-EPS)
- Towards a Ensemble Kalman filter (EnKF)
- Conclusion
- Implementation on other french airports
3A local approach
- Fog is a local phenomenon, which evolves at small
time scales - 3D fog modelling
- Horizontal and vertical resolution are too coarse
- Physical parametrization often not accurate
enough - 1D fog modelling
- Allows better vertical resolution and more
elaborate physical parametrisations - Possibility of using data from a specific
observation system - Allows high frequency of assimilation/simulation
cycles (every hour for COBEL-ISBA)
4Description of COBEL-ISBA
- Main features of COBEL-ISBA (Météo-France,
LA-UPS) - Coupling of an atmospheric model (COBEL) and of a
surface-ground scheme (ISBA) - High vertical resolution (30 levels from 0.5m to
1360m, 20 under 200m) - 1DVar assimilation scheme with site-specific
observations - Detailed physical parametrisations for fog
modelling
References Bergot et al. (2005), Weather and
Forecasting
5Site-specific observation system
- COBEL-ISBA is in currently operationnal at the
Paris-CdG airport - Specific observation system consists of
- 30 meter tower temperature and humidity
- Soil of temperature and water content measurement
- Shortwave and longwave radiative fluxes at 2 and
45m - If fog and low clouds are detected
- Height of the fog layer is deduced from radiative
fluxes at 2 and 45m - Initial atmospheric profiles are adjusted
6Evaluation of the forecasting system
- For CdG Intl Airport, Low Visibility (LVP)
conditions if - Ceiling is below 60m
- And/or
- Visibility is below 600m
- Under LVP conditions, air traffic can be reduced
half time (Paris-Cdg airport) - Evaluation with Hit Rate (HR), False Alarm Rate
(FAR)
7OPER score of the model (CdG)
- COBEL-ISBA is a useful tool for forecasters at
Paris-CdG - Local observations are very important
8OPER onset and burnoff (Paris-CdG)
- Onset time is overall too late
- Bimodal distribution for burnoff time error
- Possible influence of aerosols?
9OPER reaching the forecaster
- Intranet
- Adequate when many airports to manage
10Research objectives
- Develop an ensemble
- Probabilist forecast
- Assess the quality of the forecast
- Work on assimilation scheme
- Improve the initial conditions
- Init with fewer observations
11Research Local Ensemble Prediction System (LEPS)
- 54-members ensemble
- Perturbations based on standard deviation of
initial conditions and mesoscale forcings - Calibration of the ensemble by a bayesian method
and a training dataset - Evaluation with Brier Score (BS)
References Roquelaure and Bergot (2008), J.
of Applied Met. And Clim.
12LEPS comparison with operational bulletins
- LVP forecast for 4 period of time 0-30min,
30min-1h, 1-2h, 2-3h - 4 forecast classes certain, probable, unlikely,
ruled-out - Night of 16-17/01/2005 low stratus
6am Bulletin
T1 T2 T3 T4
Déterministic
Probabilistic
13Research on the assimilation scheme
- Goal to optimize the use of local observations
- Initial conditions for temperature and humidity
given by - B is the model error variance/covariance matrix
- Fixed 1DVAR
- Computed with an ensemble Ensemble Kalman
Filter (EnKF)
14Research estimate of B
- Computed with an ensemble
- B matrix for temperature for different init time,
mean over the 2004-2005 winter
- Important diurnal cycle ground for development
of EnKF
15Resarch ensemble Kalman Filter
- On the flow B matrixes
- 32-members ensemble
- Simulated observations over 15 days
- Results more or less identical as compared with
fixed B.
Fixed B
Forecast time
EnKF
16Conclusion
- COBEL-ISBA is a useful tool for fog prediction in
the airports - Accurate initial conditions are very important
for fog modelling - Future implementation
- Lyon-StExupery airport
- Paris-Orly airport