\griculture, 
llite remote 
n of mango 
enth Asian 
r 1990. 
production 
tely sensed 
. NNRMS, 
45. 
ory and Soil 
. case study 
1991. Crop 
nce 61, pp. 
igh Density 
vers. Digest, 
T., Natan 
Chalam, M. 
on. Abstract 
for Natural 
1t, Pune, pp. 
U.P., pp. 1- 
M. (1998). 
ation Types 
Sensing and 
yy ISRS and 
roaches for 
) pp 9-12 
IAPRS & SIS, Vol.34, Part 7, "Resource and Environmental Monitoring". Hyderabad, India,2002 
  
SURFACE INSOLATION RETRIEVAL FROM GEOSTATIONARY SENSOR 
B. K. Bhattacharya , M. R. Pandya, A.S.Nain and V. K. Dadhwal* 
Crop Inventory and Modelling Division 
Agricultural Resources Group, RESA 
Space Applications Centre (ISRO) 380015, India 
* Commission VII, Working Group VII/6 
KEYWORDS: METEOSAT, Atmospheric precipitable water, Atmospheric turbidity, cloud optical properties 
ABSTRACT: 
A study was carried out to retrieve and validate surface insolation under clear and cloudy sky conditions by a physical approach 
using visible and thermal band data from geostationary satellite over Indian sub-continent. Meteosat data of Indian ocean coverage are 
acquired for five acquisitions at 7:00, 10:00, 12:00, 14:00 and 17:00hrs. IST during 1st to 30th December 1998. A total of 150 images 
per band were analysed to produce daily and five day average total insolation output over India. Sensitivity analysis of insolation for 
different atmospheric input parameters, total precipitable water, Angstrom turbidity parameters was carried out.The retrieved 
insolation when compared to estimates from BSS measurements over six Indian stations (Hissar, Karnal, Pune, Pantnagar, Faizabad 
and Patancheru) indicated a mean absolute bias (MAB) of 1.89 and 1.76 MJm? for daily and five day average insolation estimates 
respectively. 
1. INTRODUCTION 
Insolation received at ground surface is one of the important 
inputs to crop simulation models for estimating regional NPP 
and study on energy and water balance, in a number of 
disciplines covering agro- forest ecology and global climatic 
change studies. Before reaching the ground, the solar radiation 
undergoes absorption and scattering processes with the 
atmospheric constituents such as: ozone, water vapor, mixed 
gases and aerosol. The downward solar radiation reaching the 
ground is composed of two components, a diffuse radiation 
component from the sky dome after scattering and absorption in 
atmosphere and direct radiation reaching earth with least 
interference by the atmospheric constituents. Ground-based 
measurements of insolation are generally made with 
pyranometers located at weather stations and agricultural 
experimental stations. These stations are too sparse to provide 
representative spatial variation of insolation from conventional 
meteorological observations such as: Cloud cover, cloud type, 
cloud height, bright sunshine hours (BSS) precipitable water and 
temperature. 
The feasibility of deriving and mapping surface insolation from 
satellite data has been investigated elsewhere using a number of 
approaches including regression based (Brakke and Kanemasu, 
1981), numerical (Gautier et al., 1980), physical (Rigollier and 
Wald, 1999 & 2000; Tanahashi et al., 2001) and DEM based 
(Wang et al, 2000) techniques tested with geostationary 
satellites such as: Meteosat, Europe and VISSR GMS-5, Japan, 
GOES, NASA as well as polar orbiting satellites such as: 
Landsat TM and NOAA AVHRR data. A preliminary study 
with limited clear sky Meteosat data (Bhattacharya et al., 2002) 
showed that the surface insolation could be retrieved with higher 
accuracy over Indian sub-continent using physical approach 
  
* Corresponding author: bkbbhatt@yahoo.com 
than regression and semi-physical approaches. Among the 
different physial models used in the recent past, the algorithm 
used by Tanahashi ef al. (2001) has some advantages over 
others. It takes account of atmospheric correction in great 
details. Both direct and diffuse components of insolation can be 
derived using this model. The present study is aimed at 
retrieving and validating surface insolation using Meteosat data 
of Indian Ocean coverage for both clear and cloudy skies using a 
physical approach. 
2. STUDY AREA AND DATASETS 
The Indian Ocean coverage of METOSAT-5 images both in 
visible (2.5kmX2.5km) and thermal (5km X 5km) bands were 
acquired from EUMETSAT covering 7 °- 38° N to 68-100 E 
(1338 line X 1442 pixels) for the period I" to 30" December 
1998. Time of acquisitions were 7.00, 10.00, 12.00, 14.00 and 
17.00 hrs. IST. In all, a total of 150 images per band were 
acquired for analysis. All Meteosat images were coregistered 
with INSAT 2E CCD visible band. The daily total insolation 
derived from Bright Sunshine Hours (BSS) method for the 
corresponding period were obtained for validation purposes 
over six stations in India. These stations are: Hissar (29°10'N, 
75%4'E), Karnal (29°42'N, 77°02'E), Pune (18°32'N, 73%51’E), 
Pantnagar (29°N, 79°50/E), Faizabad (26°15/N, 82°08'E) and 
ICRISAT, Patancheru (77°21'N, 78°01'E). The image processing 
and analysis were carried out using ENVI 3.4 version image 
processing software and IDL 5.3 programming language.