I too have just come across this issue when working with data touched by HEC-RAS, and, perhaps not coincidentally, with California dam failure inundation maps. But I do wonder if this is actually stemming from incorrect projection definitions bundled with ArcGIS.
Consider a similar case: the state of Wisconsin's preferred projection. It's right above California's Teale Albers projections in the coordinate system list in ArcGIS. Have a look at the False Easting and False Northing values for:
- NAD 1983 (2011) Wisconsin TM (Meters)
- NAD 1983 (2011) Wisconsin TM (US Feet)
Click between the two of them several times. You will see that the False Easting and False Northing parameters are different between them. I didn't break out a calculator to verify but they sure look like they're different by a factor of, oh, 3.28 or so.
Now go down the list and have a look at the False Easting and False Northing parameters for:
- NAD 1983 California (Teale) Albers (Meters)
- NAD 1983 California (Teale) Albers (US Feet)
Again, click between the two of them several times. You will see that the False Easting and False Northing parameters are the same between them. (Well, see the easting is 0, so of course multiplying it by any factor would make it be the same, but the northing is also the same between them, a nice round number, minus four million).
So apparently the False Easting is supposed to be specified in the linear units of the projection. Somebody at either ESRI or the US Army Corps of Engineers (the folks behind HEC-RAS), or the European Petroleum Survey Group (EPSG) misunderstood the measurement units required for these parameters in the projection.