In recent years, small-area estimation methods have received attention. This attention has been due to the increasing demand for reliable estimates for small areas, as these estimates are used for development planning, government funding allocation, and business decisions. A key question in small-area estimation is how to obtain reliable estimates when the sample size is small? When the sample size obtained from an area is small (or even zero), direct estimators have an unacceptably large standard deviation. One way to improve these estimates is to borrow power from existing data sources. To do this, we can improve the effect of sample size through modeling, or in other words, using coupling models (including explicit and implicit models). To obtain a valid estimate of the small area mean, generalized linear mixed models and the best empirical linear unbiased predictor have been widely used. In this paper, we first introduce the small area estimate. Then, to obtain valid estimates for small areas, we introduce the Phi-Heriot model (Phi-Heriot, 1979), a special case of generalized linear mixed models. Finally, in a simulation study using data from the 2003 agricultural census, orange production in the counties of Fars province (small areas) in 2003 is estimated based on the Phi-Heriot 1979 model.