For many environmental applications, the interpretation of fiber-optic Raman distributed temperature sensing (FO-DTS) measurements is strongly dependent on the spatial resolution of measurements, especially when the objective is to detect temperature variations over small scales. Here, we propose to compare three different and complementary methods to estimate, in practice, the "effective" spatial resolution of DTS measurements: The classical "90% step change" method, the correlation length estimated from experimental semivariograms, and the derivative method. The three methods were applied using FO-DTS measurements achieved during sandbox experiments using two DTS units having different spatial resolutions. Results show that the value of the spatial resolution estimated using a step change depends on both the effective spatial resolution of the DTS unit and on heat conduction induced by the high thermal conductivity of the cable. The correlation length method provides an estimate much closer to the value provided by the manufacturers, representative of the effective spatial resolutions along cable sections where temperature gradients are small or negligible. Thirdly, the application of the derivative method allows for verifying the representativeness of DTS measurements all along the cable, by localizing sections where measurements are representative of the effective temperature. We finally show that DTS measurements could be validated in sandbox experiments, when using devices with finer spatial resolution.
Keywords: active-DTS method; fiber-optic distributed temperature sensing; laboratory experiment; spatial resolution.