The 2009 Lunar CRater Observation and Sensing Satellite (LCROSS) impact mission detected water ice absorption using spectroscopic observations of the impact-generated debris plume taken by the Shepherding Spacecraft, confirming an existing hypothesis regarding the existence of water ice in permanently shadowed regions within Cabeus crater. Ground-based observations in support of the mission were able to further constrain the mass of the debris plume and the concentration of the water ice ejected during the impact. In this work, we explore additional constraints on the initial conditions of the pre-impact lunar sediment required in order to produce a plume model that is consistent with the ground-based observations. We match the observed debris plume lightcurve using a layer of dirty ice with an ice concentration that increases with depth, a layer of pure regolith, and a layer of material at about 6 meters below the lunar surface that would otherwise have been visible in the plume but has a high enough tensile strength to resist excavation. Among a few possible materials, a mixture of regolith and ice with a sufficiently high ice concentration could plausibly produce such a behavior. The vertical albedo profiles used in the best fit model allows us to calculate a pre-impact mass of water ice within Cabeus crater of 5 ± 3.0 × 10 11 kg and a mass concentration of water in the lunar sediment of 8.2 ± 0.001 %wt, assuming a water ice albedo of 0.8 and a lunar regolith density of 1.5 g cm −3 , or a mass concentration of water of 4.3 ± 0.01 %wt, assuming a lunar regolith density of 3.0. These models fit to ground-based observations result in derived masses of regolith and water ice within the debris plume that are consistent with in situ measurements, with a model debris plume ice mass of 108 kg.