A four replicated maize (Zea mays L.) crop irrigation management experiment was carried out in a Typic Haphortox soil of Guaíra (SP), Brazil (20°27'30''S, 48°19'30''W, 495 m) from July to December, 1994. There were three treatments: tr000 - half of the recommended amount of water was applied during the crop cycle; tr0I0 - the recommended amount of water was applied only during the intermediate stage and half of it during the initial and the final stages; and trIII - the recommended amount of water was applied during the whole period of the crop. The soil water matric potential was assessed by means of tensiometers, from 0.10 to 1.00 m deep, at each 0.10 m to study the variability propagation of matric potential to the soil water content. Ranges of matric potential and soil water content values, with high variability, were also identified. The results showed that: (1) occasionally, measurements of matric potential and soil water content just after irrigation or rain showed a considerable increase of variability due to the fact that in some replications, the wetting front did not reach the tensiometer porous cup; (2) the matric potential and its stability over the time increased with depth, but its spatial variability decreased at depth; (3) the matric potential variability was high and showed variance heterogeneity; (4) the variability of soil water content was less accentuated than that of the matric potential for the whole studied range. According to the results, it was concluded that the irrigation management can be done with a high level of statistical confidence when matric potential values are converted into soil water content values, because the variability problems are minimized by this transformation.
tensiometer; retention curve; error propagation