Improving Spatial Resolution of Raman DTS Using Total Variation Deconvolution

Traditional Raman distributed optical sensor (DTS) exhibits a low-pass characteristic that causes sharp temperature changes to be over smoothed. This behavior can be modeled as the convolution of the real temperature profile with the DTS impulse response. This paper presents a deconvolution algorithm developed to improve the spatial resolution of a Raman DTS system. The algorithm is based on a linear DTS model and total variation regularization. The main advantage is the ability to correctly reconstruct hot regions with dimensions down to 15 cm, which represents a resolution gain of up to six times when compared with the DTS spatial resolution of 1 m. We present simulations and experimental results demonstrating the efficacy of the proposed method.