Aucun titre de diapositive

1

• Comparisons (April - August 2003)
• Model predictions
• Diagnostic using model outputs
• Impact of temperature and humidity biases on the
  surface fields
• Turbulent and radiative fluxes at the surface
• Perspectives

2
Selection of days between April and August
2003 Cabauw 95 days Chilbolton 81days Palaiseau
75 days Models ARPEGE IFS Met-Office model
turbulent fluxes are not available RACMO
results are strange more test are
needed Comparisons between models and
observations done on an hourly basis
3

• Frequency distribution of the difference between
• observed cloud base height
• predicted cloud base height

• Large proportion of disagreement
• ARPEGE under-estimates the cloud base height
  (better at Chilbolton)

4

• Possible sources of error
• Are the ground-based observations representative
  of what is happening on a larger spatial scale
  (model grid point) ?
• Wrong prediction of the large scale
  meteorological situation ?
• Boundary layer scheme ?
• Cloud scheme ?
• diagnostic computation
• Physical parameters retrieval using model outputs

5
The condensation level of a parcel of air coming
from the surface cloud base height

• Frequency distribution of the difference between
• observed cloud base height
• condensation level of a parcel of air observed
  at 20m high

• Observed clouds are predominantly created via
  surface driven processes

6

• Advantage of working with diagnostics
• Comparisons between models are independent of
  the parameterizations

7

• Slightly better overall agreement than with the
  model predicted CLBH
• Essentially same flaws than the predicted CLBH

8

• Observed clouds are essentially created via
  surface driven processes
• Surface driven processes diagnosed using model
  outputs do not create right clouds
• ? Model bias near the surface at the origin of
  the wrong cloud base height prevision.

9

• ARPEGE over-estimates T
• ECMWF slightly under-estimates T

10

• strong variability in RH
• ARPEGE over-estimates RH

11
CABAUW
SIRTA
12
What could be the error sources to explain
surface biases ?

• Assimilation ?
• Soil scheme ?
• Boundary layer scheme ?
• Radiative and Cloud scheme ?

13
CABAUW
SIRTA
ARPEGE new radiative and cloud schemes (during
2 month)

• ARPEGE LW fluxes are under-estimated ?
• IFS no clear tendencies

14
CABAUW

• ARPEGE under-estimates the latent and sensible
  fluxes which does not explain the temperature and
  relative humidity biases

15

• For selected days of cloudy convective boundary
  layer on the CLOUDNET stations
• Predicted boundary layer cloud base height
  further than 300m from observations
• 40 of the hours for IFS
• 55 of the hours for ARPEGE.
• Same behavior in the different stations.
• ARPEGE
• Under-estimation of the CLBH due to warm and
  humid biases at the surface
• Essential condition to have a good prediction of
  dry and cloudy boundary layer diurnal cycle
  right surface field prediction.
• Soil scheme
• Surface layer scheme
• Precipitations (convection)