I just wanted to finish this up, with a general recipe for doing it right.
The paper I was quoting was basically saying that you didn't need very expensive accelerometers if you had a broadband (broad range) seismometer. This is basically true since modern seismometers are capacitive accelerometers with a larger proof mass. They just directly integrate to velocity. Cheap accelerometers are mass-produced chips used for other purposes, and make fine limited-range accelerometers. More expensive, custom accelerometers use small-batch etched and micro-machined masses that are still quite small. The installation at Darlington used such instruments, and they are very rugged.
Still, cheap and wide deployment has it's benefits, as we have discovered from Chile. In my neck of the woods, I would pay more attention to calibration, and there lies the rub. The seismometers I helped deploy became calibrated by recording regional earthquakes. There is always the chance that you put the leads in backwards, or you located on a bad site. You can tell this by comparing the results to all the other local seismometers, and seeing that the new one generally fits. You can actually get a very precise account of the local site amplification.
Calibrating accelerometers on varying site conditions is a bitch. Chile appeared to have uniform soil conditions and all the (very old) accelerometers seemed to agree with each other. For NA I would make a special effort of calibration by purchasing the expensive broad-range accelerometers, and recording them continuously on a buffer, connected to the Internet. I would never use a trigger, and would fetch back the results, based on local seismometer detection of events. I've had lots of good earthquakes which should have been recorded by the accelerometers, but never the triggered the systems. This is endemic in NA.
Seismometers have a serious problem in noisy urban environments, but we can have some success at night, to calibrate the system. We have a seismometer on the Leslie St. Spit in Toronto, which is totally blanked during the day by noise, but is very sensitive at night. Nearly all other sites around Toronto have been rejected because of the severe noise, probably because the city is on the hanging wall of the Western Lake Ontario fault.