International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
  
temperatures through relationships determined using the 
TRMM PR precipitation products as reference. Evaluation of 
the algorithm (defined as the Sferics Infrared Rainfall 
Technique, SIRT) performance based on daily rainfall rate from 
Global Precipitation Climatology Project (GPCP) is done in a 
qualitative way. 
2. DATA PROCESSING AND EXPLANATION 
ALGORITHM 
The initial idea was to use simultaneous infrared GOES data, 
lightning data and rainfall measured by meteorological RADAR 
onboard on TRMM over Paraíba do Sul basin area (figure 1a), 
for the period from September of 1999 to March of 2000. 
However it was not possible due to amount of data to establish 
a consistent statistics curves for rainfall estimation for the study 
area. The latter has been solved by extend the training area as 
shown in figurelb. 
  
  
      
  
  
  
  
  
EY p d ci Rey i eem mm 
ORCI 
3 CE m és 2156 n x > is EC 
& A TN AN : ^ 
tv ; oA i 
10 \ Mops / à y is 
rr : A 
30 4f. RO. / 3 
/ | SPALT. 
bo... -80. m $6. bo ifs 
  
  
  
  
  
(a) 
(b) 
Figure 1. Lightning space distribution represented by the dark 
areas: (a) Paraiba do sul river area (b) extended area. 
The GOES brightness temperature, lightning data and rain 
(TRMM) were project and interpolated in equidistant projection 
of 0.1 x 0.1 degrees (~ 10 km x 10 km). 
For evaluation of the effectiveness of the methodology for 
rainfall estimation via infrared GOES data in this work, a set of 
4738 infrared GOES data and 1621 rainfall from TRMM-PR 
data were used. The table 2 shows the amount of lightning and 
lightning free cloud simultaneously used to establish the curves 
Table 2. Cloud number with lightning (LTG) and lightning free 
(NLTG) simultaneously collected with GOES infrared data and 
rainfall from TRMM-PR. 
  
Type of Clouds Number of Clouds 
  
Lightning (LTG) 71 
  
  
  
Lightning free (NLTG) 693 
  
  
  
The algorithm is based on the following hypotheses: (1) 
lightning information can advance our ability to discriminate 
convective rain areas within a cloud system, given that current 
IR algorithms face limitations in dealing with the presence of 
cirrus clouds and non-raining cloud shields like those in 
Mesoscale Convective System (MCS); and (2) the development 
of a combined lightning and satellite IR algorithm based on the 
newly available PR rain products can lead to improvements in 
the definition of rain area, precipitation classification, and 
quantitative rain estimation from these sensors. The algorithm is 
designed to produce maps of instantaneous surface rainfall 
fluxes and rain type classification based on a combination of 
lightning, IR brightness temperature (10.2-11.2 um) 
observations, and precipitation profile measurements from PR. 
Besides the algorithms classify clouds with lightning (LTG) and 
lightning free (LTG), this classifies the clouds system on the 
land or ocean, and if is a stratiform or convective one. This way 
the following calibration curves are generated: Brightness 
temperature distribution and rainfall rate, rain area and 
convective fraction, and assignment of rainfall rate. 
2.1. Brightness Temperature and Rainfall Rate 
The figure 3 presents the temperature distributions for the rain 
clouds with LTG and NLTG over the land and ocean. It can be 
noticed that the clouds that possess LTG present a larger 
frequency of colder tops, in other words, larger vertical 
  
  
  
  
  
  
  
  
development. The convective area also presents colder 
temperatures. 
Convective - Land Stratiform - Land 
95 7 25 
| 
2.07 20[ à | 1 
c 5 ! 
P 4p ? pl 
Z 15 i1 M^ | 
0.5¢ i 0.51 
OO ul 0.0 
180 196 212 228 243 259 275 180 196 212 228 243 259 275 
cm 
7e 
Freq(% 
Freq{x) 
  
80 196 212 228 243 759 275 
TK) 
  
Figure 3. Distribution of frequencies of brightness temperature 
for rain clouds with lightning (red) and lightning free (blue) for 
convective (left) and stratiform areas (right) over the land (top) 
and ocean (below). 
The figure 4 illustrates the distribution of rainfall rate for the 
rain clouds with LTG and NLTG. Differently of the 
temperature distribution, it doesn't happen a significant 
difference for the rainfall rate for the clouds with LTG and 
NLTG, as presented by Morales and Anagnostou (2003). 
1258