The Latin American Giant Observatory (LAGO) consists of a network of water Cherenkov detectors (WCDs) installed in the Latin American region at various latitudes, from Sierra Negra in Mexico, 18°59′ N 97°18′ W to the Antarctic Peninsula, 64°14′ S 56°38′ W and altitudes from Lima, Peru at 20m a.s.l. to Chacaltaya, Bolivia at 5500m a.s.l. The detectors of the network are built from commercial water tanks, so they have several geometries (cylindrical in general) and different water purification methods. All these features generate different profiles in the response to air shower particles measured by our detectors and produce pulse-shaped electronic signals. Common sources of noise in a WCD come from light leakage, electronic noise, and noise associated with the operation of photomultiplier tubes (PMTs) such as thermionic emission and after-pulses; they all could produce detectable pulses recorded by the LAGO data acquisition (DAQ) system. In LAGO WCDs, these noise signals are expected to present a short pulse width (of a few nanoseconds), while secondary radiation typically produces pulses of several tens of nanoseconds. We used data from the LAGO DAQ system, which digitises pulses at 40MHz sampling rate on windows of 300ns (12 temporal bins) and with a 10-bit resolution. The LAGO DAQ configuration uses a single threshold-based trigger in the third temporal bin. We proposed a secondary trigger threshold at the fourth bin to improve the noise rejection. In this work, we show how the optimal values for these triggers are now obtained from the measurement of the muon lifetime within the water volume and the resulting Michel spectrum. Our results were also simulated using the LAGO ARTI simulation framework to estimate the expected flux of secondary particles at the detector site; and the Meiga framework, a Geant4-based simulator used to estimate the WCDs response to the atmospheric radiation flux.