To get the LiDAR in I had to create a new Elevation dataset. What I did was using a terrain feature dataset loaded the LiDAR points. Then using the USGS DEM, I converted each pixel to a point which then was also loaded into the feature dataset (this will mess up the point spacing, but keep it at the LiDAR spacing to let the calculations work for the areas you really need). I also had an older LiDAR collection for one of my watersheds which went outside the county so that one went into the dataset. The key was keeping the same datum and units with the different points. Then I let ArcGIS calculate away to create the terrain and then convert that terrain to a few raster in different resolutions. Hazus did have some issues with the full 5ft per pixel resolution so I mainly use a 10ft per pixel one for my analysis. The areas that were only the USGS DEM were resampled to the smaller pixel size, which while a distortion of the accuracy, but allowed me to get a rough flow into the LiDAR areas. The key areas had the good coverage and thus were able to calculate the correct flows. So far, luckily, I have only one set of flooding ground truth points (only one flood event in 3 years), and the largest difference for the edge of flooding was 1.2 feet(vertical) on the 10ft pixel with the LiDAR accuracy being .5ft.