Yield prediction modeling of rice crop was carried out using
the collected data from Sakha experiment station, Agriculture
Researcher Center, Ministry of Agriculture, Egypt. The experimental field was situated in the rice belt region which includes
Kafr El-Sheikh Governorate. It is located between 31 060 4000
and 31 060 000 North and 30 540 3000 and 30 550 6000 East
(Fig. 1). The total area of rice observation site was 2.4 ha during
the growing seasons of 2008 and 2009, cultivated by the variety
Sakha 104. The region that includes the study area is defined as
Pro-Deltaic Alluvial Plain. This Pro-Delta is characterized by
clayey soil of high clay fraction and high water saturation percentages. These clayey soils are characterized as Vertisols of Typic Haplotonerts, fine, and thermic (Afify et al., 2011). Rice was
sown in May 24th and 23rd in the 1st and 2nd seasons, respectively. At 90 days from sowing (maximum vegetative growth
stage), sixty measurements were collected from sixty parcels
of the rice field in each season based on the grid system
(Fig. 2). Each parcel covers 400 m2 (20 · 20 m) that represents
a single SPOT pixel that was fixed as one plot of measurements.
The location of the center square meter of each plot was recorded using global positioning system (GPS). Out of this number, fifty random samples were selected for the modeling process
and ten samples were selected for validation. Three types of data
were used as inputs for generating rice yield prediction models:
the direct spectral data collected from SPOT imagery (reflectance values of green, red and near infrared bands), six calculated vegetation indices values, as well as the values of
observed rice yield and LAI.