The ELDORADO surface radiation system

1
The ELDORADO surface radiation system

• Bart vd Hurk Dirk Meetschen et al
• (see ELDAS web-page)

2
Principle

• Generating 3-hrly LW SW surface downward
  radiative fluxes
• ELDAS grid resolution
• Combining METEOSAT data and RT calculations from
  the ECMWF model
• Simplified data assimilation technique
• Brief verification using routine SW stations in
  The Netherlands

3
ELDORADO steps (1)

• Create initial field of cloud cover liquid
  water profile
• Continuous simulation using LAM version of ECMWF
  model at 0.5?, started daily from interpolated
  ECMWF analysis
• Interpolate cloud fields at 3hrly intervals to
  ELDAS 0.2? grid
• Create METEOSAT-derived cloud cover field at
  ELDAS grid
• METCLOCK cloud mask product
• Cloud cover average mask of 5?3 METEOSAT pixels
  surrounding target location

4
ELDORADO steps (2)

• Create METEOSAT net TOA SW radiation at ELDAS
  resolution
• Sensor irradiance ? net TOA flux
• Extra time/space/month-dependent calibration
  factor, derived from monthly averaged clear sky
  TOA SW flux generated by LAM

6UTC 9UTC 12UTC
5
ELDORADO steps (3)

• Reshuffle model columns for relocation of
  cloudy/cloud-free atmospheric columns
• For each target location, select model column
  with the smallest value of
• J a distance b ?albedo c ?SWnetTOA
• Adjust total cloud cover
• to METCLOCK value
• by scaling cloud
• cover profile

MetClock obs
height
scaling factor
First guess Scaled
Cumulative cloud cover
6
ELDORADO steps (4)

• Adjust liquid water content similarly in order to
  match SWnetTOA
• Non-linear relation 2 iterations applied
• Only applied to cloudy model gridpoints
• Store and interpolate resulting downward LW and
  SW surface fluxes as best estimate
• When either METCLOCK or SW-obs are absent use
  first guess
• Archived METEOSAT/METCLOCK data covers only part
  of the ELDAS domain

7
Total cloud cover
8
TOA SWnet
9
Surface verification

• ? 30 Routine stations in Netherlands
• Only downward shortwave
• Daily totals
• First guess affected by erroneous daily reset to
  cloudfree conditions at 0 UTC probably little
  cloud spin-up in morning hours

10

• Statistics
• April/May 2000
• this dataset
• first guess 37 W/m2
• ELDORADO 12 W/m2
• Bias of ERA40 over
• SGP ARM observations
• (April-May 1999)
• clear sky 31 W/m2
• overcast 26 W/m2
• all cond. 17 W/m2

11
Still to do

• Further validation
• BSRN stations (hardly available via www-service,
  but Lindenberg-data are ordered)
• Validation plan
• Cleaning the archive
• Replace missing time slots by first guess
• Fill area outside METEOSAT domain with first
  guess
• Interpolation to ELDAS grid