The ice sheet-ocean modeling community is making large strides toward developing coupled models capable of examining the interactions and feedbacks between ice shelves and ocean along the Antarctic margin. We present preliminary results and address some of the challenges that have arisen during the development of a coupled ice sheet-ocean model. The ice sheet model is icepack, a shallow-shelf finite element model written in Python. The ocean model is the Regional Ocean Modelling System (ROMS), a terrain-following vertical (sigma) coordinate model that has been modified to interface with a moving ice shelf. These two models are coupled in an online configuration using the Framework for Ice Sheet Ocean Coupling (FISOC). The use of a model with sigma coordinates for the ocean component introduces a simplification and a complication to modeling a moving ice draft. The sigma coordinate system retains the same number of vertical layers at any depth, eliminating the need to convert grid cells between ice and water, when using a fixed grounding line configuration. However, as the ice shelf draft evolves in time, topographic configurations develop that induce pressure gradient errors in ROMS. We quantify these errors in an idealized set-up with an artificially changing ice draft following the ISOMIP+ geometry. We compare results between an ice draft that is smoothed to meet standard ROMS smoothing criteria (rx0, rx1) and a non-smoothed ice draft. Finally, we present a simple parameterization in a buffer zone near the grounding line that uses interpolated melt rates from the ocean model, allowing us to maintain a steep ice topography in the ice model without inducing pressure gradient errors in the short water column in the ocean model. This model configuration will be applied to Pine Island Glacier and used to examine present and possible future states of the ice sheet-ocean system.