Distributed temperature sensing (DTS) was used to record temperature from early 2016 to present for a Marcellus Shale horizontal dry gas well, MIP-3H, located in Monongalia County, West Virginia. In addition, after wellbore clean-out with water and nitrogen a flow scanner production log was conveyed on March 02, 2017. The flow scanner provides one day of gas and water production from each of the 28 stages in MIP-3H and from each of the clusters. The DTS data provides an opportunity to inspect the reservoir for Joule-Thompson (JT) effect, a phenomenon that describes cooling of an non-ideal gas as it expands from high pressure to low pressure, and obtain a relative production attribute along the lateral of the MIP-3H. The original fiber-optic DTS data shows the temperature along the lateral; however, due to the geometry of the well with toe up and the presence of a small fault and minor water production at Stage 10 relative gas production of each stage cannot be directly determined from the raw DTS data. We present two methods to generate DTS attributes that can be used to better reveal relative gas and water production through time from each perforation cluster and each stage of the MIP-3H. The first attribute deals with the deviations of the DTS measurements from the calculated geothermal temperature, while the second attribute calculated the difference between DTS temperature and the average daily DTS temperature along the lateral of the MIP-3H. We show that the latter DTS attribute provides a more robust image of temperature variations regime along the lateral than the former attribute. Negative values of the DTS attributes reveals JT cooling, resulting from stages of the MIP-3H with higher natural gas production. A correlation analysis of the production log with the calculated DTS attributes suggests that the production log is not representative of the entire production life of MIP-3H well. Temporal correlation with the DTS attributes is highest close to the production log recording day (March 2, 2017) decrease rapidly and the weak correlation switches from positive to negative.