The time domain reflectometry (TDR) technique has become a popular way of measuring soil water content and the use of multilevel probes increases its applicability. The few existent studies about multilevel probes in the field suggest their use could be problematic. An experiment with 48 five-segment probes was conducted in a transect in a Red-yellow Oxisol during two periods in 2000 and 2001, collecting data of soil water content by tensiometry simultaneously with the respective TDR measurement. Detailed soil water retention curves were obtained to allow use of tensiometer observations as a standard. Analysis showed that TDR response pattern was different between periods contrary to soil water content. It was not possible to obtain a calibration curve that could be used for all probes. Adjustments per probe or layer showed very small determination indexes. Calibration curves fitted to the data per segment showed high variability, with coefficient of determination between close to zero and above 0.98, and no trends were found in these variations. Calibration curves from the first period were not correlated to those from the second. Generally variations were smaller and determination cofficientes higher for the first period. Spatial dependence appeared to exist between calibration curves at some depths, but it could not be modeled. No correlation was found between calibration and some soil attributes. These analyses were difficult due to the large variability found for the relation between TDR response and soil water content, suggesting this kind of TDR probe is unsuitable for field applications.
soil water content; field calibration; tensiometry
