Blitzortung added a TOA detection technique to their Blue detection system.
This statement is Incorrect. Blitzortung uses both TOA and TOGA algorithms in both BLUE and previous RED SYSTEMS, and has used GPS since it's inception with the original Green systems. Private algorithms determine when and what specific TOGA zero crossing computation is employed on a specific channel, and how many iterations of the predicated algorithm to use... typically a minimum of 4. Beyond 4 iterations does not typically improve location characteristics enough to be of value, in most cases, but they can certainly be computed and processed. NASA only uses 15 points of Pi for computing orbits.
Once a stroke has been presumed to have occurred, from any station, all station signals from all channels, are processed for 'best data / quality" and primary, secondary detectors are assigned for that impulse, and it's associated pulses and characteristics. Only the 'best' channel data will be used, other channel data discarded or archived for some interval. Blitzortung requires a minimum of 8-18 detecting stations for validation of data in North America. This will vary by environmental conditions. Earlier today I noted the conditional detectors at 8-17. Although a TOA network can define and roughly locate a stroke using a minimum of 4 detectors, Blitzortung requires a minimum of 8 in the Americas Region to define a stroke. World regions with lower station density may require fewer 'detector's... those are called "Primary" detectors, with the remaining, referred to as 'secondary'. All other stations detecting the impulse are referred to as 'also detecting' or 'detectors'. The channel data from primaries and secondaries is examined, best channels determined, then processed against time and TOGA algorithms.
If you go here
http://sferics.us/ live signals from station 839 (Red) are being shown,(up to 5 data channels sent) at 1½ second intervals (not all signals, just one every 1½ seconds)... those 'hash marks' at the top of the signals represent TOGA Zero Crossings for each of the shown channels for that signal. That is displaying 'initial' processing by the servers, before final determination by further algorithms. Looking at the real-time 'last 60 second' map just below it, you may see that impulse plotted as a 'stroke' almost instantly... (delay was 1.2 seconds as I typed this... faster than the page is sampling the signals image above!) This is what Blitzortung means when it referes to "Real Time"... although 1.2 seconds delay won't be called 'real time' by some luddites or nit pickers. During peak periods where detections may run higher than 20-30 thousand per hour, I've seen the delay exceed 8 seconds... but the nit pickers' will just have to live with it.
If you go to
Lightningmap.org... there is a feature, which is not always active, and may vary, ... 'experimental' ,,, this is a pull down under the upper-left map controls displaying the 'time range' for the map display. While we don't always know what they're playing with there, most of the time it has to do with refining 'false positives' or 'polarity' or 'type' testing for future implementation.
As far as accuracy of location, I've personally gone out and spotted detected strikes within meters of where BT centered it's 'deviation error' circle. Others certainly within the computed 'deviation' circle. Accuracy is best for C-G types. Anvil Crawlers, other types may be miles long, never strike the ground, or may have traveled miles before actually grounding, so any system will have wider errors on types such as that.