Compensation for temperature dependence of Stokes signal and dynamic self-calibration of a Raman distributed temperature sensor

This article discusses a powerful calibration technique that compensates for the temperature dependence of the Stokes signal in a Raman distributed temperature sensor. The Stokes signal is conventionally used for normalization since its scattering cross-section is a weaker function of temperature than that of the anti-Stokes signal. The potential of this temperature dependence of the Stokes signal to introduce errors in measurements has received only cursory mention in the literature. We show that this effect degrades the temperature accuracy by as much as 90%. Additionally the heavy fiber attenuation at the operating wavelength also makes calibration non-trivial. The technique described implements a corrective mechanism for this temperature dependence, offers the additional benefit of self-referencing of a heated curve rather than referencing to the fiber at room temperature, and demonstrates the viability of such a system despite the heavy fiber attenuation at the operating wavelength. This obviates the impractical and erroneous approach of storing an initial room temperature profile for subsequent temperature calculation. This dynamic self-referencing adds substantial immunity to the tolerances in the actual numerical values of the Raman wavelengths thereby making the system robust.