Clouds METEOSAT

1
Comparison of cloud statistics from MSG with
regional climate model data
Overview Meteosat multi annual dataset METEOSAT
past 30 years, Meteosat Second Generation (MSG)
in operation, Meteosat Third Generation (MTG) in
preparation. Dataset available for of atmosphere
and surface parameters climate analysis. Cloud
detection with MSG data Based on multi channel
spectral information. Algorithms for day and
night, including single channel thresholds and
channel combinations. Atmospheric water cycle and
cloud appearance Algorithm for analysis of cloud
cover, structure and cloud types. Analysis
results can be used for model validation and
detecting possible climate changes.
Cloud detection in MSG Data
Dynamic IR 10.8µm threshold for each pixel and
slot

• 30-day maximum TOA temperature
• Fit of diurnal temperature curve with thermal
  surface parameter (TSP) model (Göttsche and
  Olesen, 2001).
• TSP model temperature is basis for IR threshold.

Fog detection night Difference in
emissivity (10.8µm 1 3.9µm 0.8) results in
BT difference.
Snow detection day At 1.6µm no reflection from
snow. Big ratio of 0.6µm/1.6µm identifies
clouds as snow.
Colour code day night, day only, night only

• Comparison with model data
• MSG cloud mask versus modelled cloud cover from
  climate version of the local model (CLM DWD)
• Model area see topography compared region red
  box
• March August 2005
• Special interest convective situations
• Agreement MSG CLM cloudy or non cloudy (per
  pixel slot)

Results convective days
Difference convective days non convective
2
3
1
Agreement on 69 convective days between MSG and
model clouds average 73. Good agreement over
Black Forest (1). Upper Rhine Valley (2) and
Swabian Alp (3) show poor agreement.
Agreement is generally worse for convective
days. The convective parameters in the CLM should
therefore be modified to improve the agreement
with satellite data. The agreement e.g. over the
Swabian Alp drops significantly on convective
days.
Topography of model area. Red box compared
region
Agreement between MSG and model clouds average
80.
Outlook In a next step the cloud analysis will
be done on segment basis rather than pixel. Size,
neighbourhood, sub-segments can give more
details on clouds. Despite a similar spectral
signature, clouds can then be assigned to
different classes. By changing the scale
parameter structures within clouds or cloud
fields can be identified.
Literature Göttsche, F. M., Olesen, F. S.
(2001). Modelling of diurnal cycles of
brightness temperature extracted from METEOSAT
data. Remote Sensing of Environment, 76, 337-348