al 
RAINFALL ESTIMATION USING SATELLITE DATA FOR PARAÍBA DO SUL BASIN 
(BRAZIL) 
Palmeira, F. L. B.*, C. A. Morales ^, G. B. Franca ©, L. Landau ? 
“ Department of Civil Engineering, COPPE, Universidade Federal do Rio de Janeiro , Rio de Janeiro, Brazil 
felipepalmeira@acd.ufrj.br, landau@lamce.ufrj.br 
? Department of Meteorology, IAG, Universidade de Sáo Paulo, Sáo Paulo, Brazil 
morales@model.iag.usp.br 
* Department of Meteorology, IGEO, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil 
gutem@acd.ufrj.br 
KEY WORDS: Meteorology, Hydrology, Precipitation, Estimation, Satellite, Algorithms, Infrared, and Temperature. 
ABSTRACT: 
This work presents a self-consistent algorithm for near real-time rainfall estimation via infrared Geostationary Operational 
Environmental Satellite (GOES) (each 4 hour) data based on a validation simultaneous dataset from Precipitation Radar (PR) 
onboard the Tropical Rainfall Measuring Mission (TRMM) satellite, GOES and lightning data, for the basin of Paraíba do Sul river. 
The dataset corresponds the period from September of 1998 to March of 2000. The rainfall estimation methodology was developed 
by Morales and Anagnostou (2003) and has been adapted for the proposed of this work and it is based on following assumptions: (1) 
lightning is correlated with the presence of precipitation and ice particles that are associated with deep convective nucleus; (2) the 
precipitation area and its convective portion are related to the cloud and lightning areas of a precipitating system; and (3) lightning 
and no-lightning clouds exhibit different precipitation characteristics. The aforementioned database was used to analyse and 
understand the main physical characteristics of the different types of raining systems which act in the study area. Results are 
presented and discussed. 
1. INTRODUCTION 
Information on the spatial and temporal variability of global 
precipitation is of fundaméntal importance to applications 
ranging from hydrologic engineering to climate change 
research. This paper presents an algorithm developed for near 
real-time retrieval of instantaneous surface rainfall over a 
region of Brazil using information from an geo-stationary 
satellite infrared observations, and adjacent rainfall 
measurements from the first space-borne precipitation radar 
(PR) onboard the Tropical Rainfall ' Measuring Mission 
(TRMM ) satellite (Theon, 1994), and lightning data. This study 
is formulated under the hypotheses that: (1) lightning is 
correlated with the presence of precipitation sized ice particles 
that are associated with deep convective cores: this observation 
can improve the identification of convective rain area; (2) the 
precipitation area and its convective portion are related to the 
cloud and lightning areas of a precipitating system; and (3) 
lightning and lightning free clouds exhibit different 
precipitation characteristics. 
The relation of lightning to precipitation has been the subject of 
various precipitation remote sensing and climatology studies. 
Workman and Reynold (1949) related flash rates to convective 
rain fluxes, and suggested that she frequency of lightning may 
be a measure of convective activity. Goodman (1990) 
developed a relationship between lightning frequency and 
rainfall intensity for systems in Florida. Similarly, Buechler et 
al. (1994) demonstrated a linear relationship between rainfall 
and lightning activity for Florida thunderstorms and Tappia et 
al. (1998) estimated convective rainfall rate from rainfall- 
lightning ratios using the Melbourne, Florida, WSR-88D radar. 
Satellite infrared (IR) images have been used to retrieve rainfall 
at large spatial and temporal scales, and for delineation of rain 
areas in cloud systems (Arkin and Meisner, 1987). Adler and 
Negri (1988) developed a technique to distinguish convective 
and stratiform precipitating systems based on the temperature 
gradients evaluated around the minimum temperature in the 
cloud clusters. Recently, Vicente et al. (1998) presented an 
auto-estimator IR technique that uses additional information of 
precipitable water and relative humidity from a numerical 
weather prediction model. These IR rainfall estimation methods 
have deficiencies associated with the presence of thin non- 
precipitating cirrus clouds and non-raining cold Mesoscale 
Convective System (MCS) cloud shields. Anagnostou et al. 
(1999) in an effort to minimize this uncertainty used a 
statistically adjusted IR technique with microwave sensors and 
showed that the area within a cloud cluster whose temperature 
is at or below the most frequent temperature in that cluster is 
well-correlated with rain area. They were also able to improve 
the convective and stratiform rain area delineation in those 
precipitating systems. Despite those efforts on improving IR 
algorithms, there is considerable uncertainty in the estimates 
since the relation between cloud-top longwave IR brightness 
temperature and the underlying surface rainfall is complex and 
is based on indirect physical relationships. Morales et al. 
(1997), and Morales and Angnostou, (2003) have shown that 
lightning measurements associated with active convection in the 
clouds can provide reliable delineation of the convective cores, 
which would lead to improvements in the convective rain 
estimation. 
This paper presents a lightning-IR-rainfall algorithm that 
consists of procedures for estimating rain area and its 
convective/stratiform portions for clouds with lightning (LTG) 
and lightning free (NLTG). Convective and stratiform rainfall 
rates are related to lightning rates and IR brightness 
1257