The increasing shortage of the renewable water resources in the country has made the farm water needs estimation to become as one of the important priorities in agricultural water management. Farm water needs are normally controlled by climatologic factors. It equals to the reference evapotranspiration which is corrected by a scaling factor associated to the crop kind and to local characteristics. In this research, using Landsat satellite imagery, we estimated and compared the crop coefficients for main agricultural crops in the Moghan cultivation industry, from two procedures: the first based on evapotranspiration measuring, and the second based on NDVI measuring. The comparisons in the case of the five main crops showed that the Root Mean Square Errors are within an acceptable range, leass than 0.28. In the following, the evapotranspiration based crop coefficient has been used in order to estimate farm water needs. Farm's water needs are indeed estimated by six methods: a combination of two actual evapotranspiration and three reference evapotranspiration ways. Among the six methods, the Metric/PenmanMonteith method was selected for final step, i.e. farm irrigation needs. The farm irrigation needs is equivalent to farm water need minus effective rain. We compared four different ways for estimating the effective rains but preferred the FAO method assigned for low slopes. Based on our results, farm irrigation needs in the Moghan cultivation industry range from 270 mm (for rainfed barley) to 1500 mm (olive groves). Statistical investigation in three years data revealed a dependency between yield performance and evapotranspiration rate. In addition, it showed that yield performance is correlated with crop spectral indices such as NDVI, LAI and SAVI. The primary goal of this research is to estimate local agricultural crop coefficient in the Moghan cultivation industry. The second goal is to investigate of relationships between crop coefficient and crop spectral indices in order to make the crop coefficient estimable directly from spectral indices.