OPB,

This has not my experience with GPS at all. We located urban buses, in some cases down to less than 50cm with GPS. We found outside tolerances, in most cases, of no more than 3 meters. Writing a proper GIS data base of the route, which is closed, and doing a combination of verification, from inertial, wayside, and GPS sensors should correct for even slightly larger variations. The use of the route data base and inertial verification is one of the way to avoid total dependence on GPS signals. This allows for the loss of satellites due to buildings or tunnels for short intervals.

One of the things we wondered, when we found several USGS markers and used these for verification of these very close tolerances was: how actually accurate were the survey markers? These of course were done by manual survey crew fifty to hundred years ago? Before we even knew about the earth's bulge?

My experience with bad GPS is because of bad GIS data bases, errors in the data inputs, and poor resolution and accuracy of the location inputs, against which the signal is being compared. Many of auto navigations systems are of this poor quality of mapping, so don't always blame the GPS.

$250k a mile may seem somewhat high, but I believe that this project also included major signal and communications upgrades? Signalization upgrade costs have skyrocted in recent years due to materials, high demand, and limited competition. This has also been the case with grade crossing and traffic signals, as well.