4. Conclusion
In the current study, after observing the results of all generated
models for rice yield prediction for the two seasons of 2008 and
2009 and the validation analysis for the generated models, it
could be concluded that using multi-regression model of LAI
as one input factor and NDVI or any other vegetation index
that is calculated from red and near infrared spectral reflectance under normal environmental conditions and common
agricultural practices during the period of the maximum
vegetative growth could be the best methodology of rice yield
forecasting using satellite imagery. Using high resolution
satellite imagery is necessary to be able to isolate rice
cultivation especially in the intensive agricultural lands of Nile
Delta in Egypt and it is also necessary to apply these models
over national scale. All generated models are empirical models
limited to environmental conditions and applicable under similar conditions.