Characterization of connectivity in compartmentalized deepwater Gulf of Mexico (GoM) reservoirs is an outstanding challenge of the industry that can significantly impact the development planning and recovery from these assets. In these deep formations, temperature gradient can be quite significant and temperature data can provide valuable information about field connectivity, vertical fluid displacement, and permeability distribution in the vertical direction. In this paper, we examine the importance of temperature data by integrating production and temperature data jointly and individually and conclude that including the temperature data in history matching of deep GoM reservoirs can increase the resolution of reservoir permeability distribution map in the vertical direction. To illustrate the importance of temperature measurements, we use a coupled heat and fluid flow transport model to predict the heat and fluid transport in the reservoir. Using this model we ran a series of data integration studies including: 1) integration of production data alone, 2) integration of temperature data alone, and 3) joint integration of production and temperature data. For data integration, we applied four algorithms: Maximum A-Posteriori (MAP), Randomized Maximum Likelihood (RML), Sparsity Regularized Reconstruction and Sparsity Regularized RML methods. The RML and Sparsity Regularized RML approaches were used because they allow for uncertainty quantification and estimation of reservoir heterogeneity at a higher resolution. We also investigated the sensitivity of temperature and production data to the distribution of permeability, which showed that while production data primarily resolved the distribution of permeability in the horizontal direction, the temperature data did not display much sensitivity to permeability in the horizontal extent of the reservoir. The results of these experiments were compelling in that they clearly illuminated the role of temperature data in enhancing the resolution of reservoir permeability maps with depth. We present several experiments that clearly illustrate and support the conclusions of this study.