Detecting transient events in noisy time series is complicated by several issues. We focus on the issue of nonstationary noise, illustrate an approach (“dynamic” strategy) on radio frequency (RF) observations from the FORTE satellite that frequently updates the estimated background, and provide a dynamic survey of FORTE noise. The performance measure is the distribution of R2/, where R2 is the squared noise amplitude at a given frequency and is the most recent mean squared amplitude. The definition of “most recent” ranges from 10 μs to 10 s. We also vary the fraction of the update period used to compute the mean from 0.1% to 100% of the previous period. For FORTE, we define 13 geographic regions and analyze signal-free records (assumed to be representative of the noise, but discussion is provided). The best dynamic case is compared to a static strategy that uses R2/static, where static is the mean over a period that covers tens of days (tens of days is the time required for the variance of the noise to reach its maximum). It is shown that some type of dynamic strategy has better statistical sensitivity than the static strategy. This survey illustrates one way to select an “update-the-mean” strategy that is applicable in many settings. We also describe a multiband triggering method, which together with the update-the-mean strategy helps establish the lower limit of detection of signals of interest, for example, in the case of FORTE, in satellite-based long-baseline radio astronomy and in cosmic ray shower RF emissions.