1303
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
with geometry and data conversion into appropriate unit). For
these procedures ILWIS is operated from the command prompt
using the ILWISComClient utility. For (near) real-time
visualization batch routines that are run using the Scheduled
Tasks Manager of the Windows Operating System initiate
ILWISComClient and ILWIS based scripts are executed that
handle the import, processing and visualization in an automated
manner. Another small batch routine, called gettime.bat
(developed by Frédéric Cazenave, Radar Engineer, Cotonou,
Benin), handles the time (zone) and the file time-stamp offset to
ensure that correct files are imported and processed. In all cases
use is made of IrfanView (http://www.irfanview.net/) for
visualization using the command line options in a batch routine.
Having efficiently imported and (pre-) processed (and
eventually visualized) the data further analysis is possible using
the generic GIS, Remote Sensing and (time-series) calculation
capability of ILWIS for different applications.
be created. It is also possible using the scroll button of the
mouse to interactively visualize an animation on the screen.
F
Figure 1. Near real-time Meteosat visualization
Next to these resources another tool was developed to directly
copy, import and visualize the Multi Sensor Precipitation
Estimate from Meteosat 9 without the use of GEONETCast
(MPE is available at: http://oiswww.eumetsat.org/SDDI/html/
grib.html), although the data is available in the data stream
(MPEG). This was done to facilitate organizations that
currently do not operate a ground receiving station but have
internet connection. For data import into ILWIS use is made of
another GRIB2 import utility
(http://www.cpc.ncep.noaa.gov/products/
wesley/wgrib2/index.html). The batch routine developed is
configured such that it checks for a new MPE file every 15
minutes at the Eumetsat website, copies the grib file(s) to a
local hard disk - directory and imports - processes the data.
Using IrfanView an animated sequence of images can be
visualized if the application is operated for a couple of hours.
The generic ILWIS functionality, e.g. the Maplist Statistic’s
Sum Function facilitates computation of the 24 hr aggregated
rainfall of the 96 initial MPE products. The described toolbox,
or parts thereof, is currently operated at the Geographic
Information Systems & Remote Sensing Research Centre
(CGIS) of the National University of Rwanda in Butare,
Rwanda, at the African Monsoon Multidisciplinary Analysis
(AMMA) project (in conjunction with IRD, France) in Benin,
University of Connecticut, USA and University of Dresden,
Germany. Also a weather amateur from Wolfsberg, Austria
reported that the toolbox was operational.
4. EXAMPLES FROM GEONETCAST USING THE
TOOLBOX
A number of examples are provided below demonstrating the
versatile capability of the toolbox. Most routines can be
operated fully automatically when included in the Windows
Scheduled Tasks Manager and facilitate image import,
calibration and visualization at user defined time intervals. At a
given system clock time a batch file is executed, a year-month-
day-hour-minute string is generated and corresponding image or
product from the GEONETCast data stream (or directly from
the Eumetsat website) is extracted and processed. The
geographic window can be easily adapted and a time stamp can
be added. Figure 1 shows a pseudo natural colour transformed
composite of channels 006, 008 and 016 (left) and a pseudo
colour transformed image using the same low resolution
channels in combination with the High Resolution Visible
channel of MSG, resampled to 1 km spatial resolution. The
output is exported to a tif and using IrfanView animations can
Figure 2 shows the result of a classification using MSG channel
108 and 062, applying a threshold on the temperature difference
of less than 11 Kelvin (an empirically determined threshold by
Kidder et al, 2005) giving an approximation of the clouds that
have a high likelihood of precipitation (left). The centre picture
shows the cloud phases using MSG channel 108 brightness
temperature (Tb); classified as cloud and Tb above 261 K, Tb
between 243 and 261 K, Tb between 233 and 243 K and Tb
below 233 K, in blue, green, orange and red respectively. These
classes represent mainly water vapour-water droplets, mixture
of water droplets and ice crystals, predominantly ice crystals
and ice cloud respectively (thresholds after Strahler, 1968, pp.
187). The left hand picture shows the clouds, classified using
the Cloud Top Height product. The height range class
thresholds are according to Strahler (1968) to assist in the
identification of clouds types based on altitude (<500 m, 500-
1500 m, 1500-3000 m, 3000-6000 m, 6000-10000 m, > 10000
m; from low to very high respectively).
> J JT
-.v.v > ■
‘Vf A-œ
_>
¡PfHpjj
#■ ' iV
■s»
¿r
Figure 2. Cloud properties
Instability indices contained in the GII product provide some
information concerning the vertical stability of the atmosphere.
Air mass parameters can be used to issue severe weather
warnings by a skilled forecaster if a certain index exceeds a
threshold (which is not a fixed value and varies from season to
season). The retrieval algorithm only works for clear sky
conditions, no instability data is inferred for cloudy pixels
(Eumetsat, 2007). Figure 3 shows an example of the K-index
(left), K values over 30 are indicative of strong convection
potential and if K values over 40 are observed severe storms are
to be expected. The resolution is 45 km/pixel (average of 15 by
15 MSG pixels). Using the combination with precipitable water
(right) has as advantage that also an idea can be obtained of
possible rainfall as more significant precipitation events are
associated with higher precipitable water values (Dostalek and
Schmit, 2001). Also on a 3 hour basis the Cloud Analysis Image
is available with a cloud type classification (bottom).
