A simple and convenient way to conduct UQ for engineering models characterized by complex physics and expensive simulations is through construction of surrogate models. However, data driven surrogates like Karhunen-Loeve expansions may require a number of evaluations of a computationally expensive high fidelity model. Multifidelity (MF) methods seek to exploit the tradeoff between these and inexpensive evaluations of low-fidelity models, where the notion of fidelity can be linked to reduced physics, coarser discretization and other factors. In our work, we take first steps towards construction of a multifidelity surrogate based on the Karhunen-Loeve expansion, and use it to conduct UQ on random field quantities from turbulent round jet simulations. The high fidelity surrogate uses the EDDES-SA simulations while the low-fidelity method is based on RANS. Results based on a small number of high fidelity training points suggest that the MF surrogate may be better and more robust compared to a single fidelity version of the same.