The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008 
1304 
Figure 3. K-index, précipitable water and cloud analysis image 
Figure 4. Multi Sensor Precipitation Estimate 
The multi sensor precipitation estimate (MPE) is directly 
retrieved from the Eumetsat Website, so even organizations not 
operating a ground receiving station can utilize this information, 
as this important parameter is affecting people in a very direct 
way. To derive the MPE use is made of the relationship 
between cloud temperature and rainfall intensity, as colder 
clouds are likely to produce more rainfall. The MPE uses a 
statistical matching algorithm in temporal and geographical 
windows to correlate the SSM/I instrument derived rain rates 
with Meteosat IR brightness temperature images. The obtained 
relationship is lateron converted to the full Meteosat-9 temporal 
resolution and a MPE product is generated each 15 minutes. 
The algorithm performs well for the tropical and subtropical 
convection areas. The relationship is based upon SSM/I - 
Meteosat co-located pixels from 40 degree North to 40 degree 
South (Heinemann et al, 2002). With the DMSP program (two 
satellites in polar-orbit array), a given location on Earth is 
revisited every 6 hours, allowing 4 brigthness temperature 
versus rain rate calibration events on a 24 hr basis. The MPE 
example presented in figure 4 is of the same day-time as those 
given for figures 1, 2 and 3. 
Figure 4: The multi sensor precipitation estimate 
Atmospheric motion vectors are available at a lower temporal 
resolution as the product is generated by applying a correlation 
algorithm to a sequence of images. By tracking of the 
movement of the cloud field or humidity structures, winds can 
be extracted. Height is determined from the infrared 
temperature and converted to pressure 
(http://www.eumetsat.int). The para-meters extracted from the 
BUFR file are longitude, latitude, pressure, wind direction, 
wind speed and temperature. The wind direction is classified 
into three classes using pressure thresholds (< 35000 Pa, 35000- 
65000 Pa and > 65000 Pa) for the high, medium and low wind 
vectors which are displayed in red, green and blue respectively. 
The approximate elevation in meter (z) is derived according to 
an equation as given by Lunde (1980) (z = - 
1/0.00001184*(ln(Pa/l01325)) assuming normal pressure at sea 
level. For visualization a predefined ILWIS MapView is used, 
containing the country boundaries and the classified wind 
direction vectors are shown, scaled according to the wind speed. 
Figure 5. Atmospheric motion vectors 
Figure 6 shows the extracted Fire product. The header lines in 
the original ascii file are removed and the space delimited 
columns are imported into an ILWIS table, which subsequently 
is transformed into a point map an visualized with the country 
boundaries. A different symbol size is used for the possible and 
probable fires given in the table. The temporal resolution of 
this product is also 15 minutes, which makes it ideal for a fire 
monitoring system.