Any irrigation action for a field management zone, which is based on images, requires a transformation into single value. Since data distribution is ab-normal in an image, using a mean value to estimate the crop coefficient (Kc), an overlaid polygon may not represent properly its water demand. Therefore, this project’s aim was to examine to which extent different statistics of potential designated values will affect an estimated Kc, and consequently affect irrigation practices.

Satellite imageries of Landsat8 and Sentinel2a were used to produce images of vegetation indices (VIs), for cotton, almond, tomato and vineyard at six sites. Four statistical measures that can be used as a potential designated value for a field polygon, were calculated: median, mode, mean ± threshold and mean within full width half maximum of histogram peak (FWHM). Their calculated VIs values were compared, and differences were transformed into Kc values based on known models. A threshold value of less than 0.1 in Kc was determined empirically as none substantial difference.

Analysis results highlighted the preferable statistical approach to produce a single designated value, which reflects more adequately a specific field attribute, independent of sensor or crop phenological stages. Using the VI of EVI indicated no sustainable difference in all tested cases, comparing the mean ± standard deviations and FWHM.