Underestimating extremes due to random uncertainty

As we attempt to infer a system's response to external driving from measurements, random errors in the measurement of the drivers can lead us to mistakenly infer a non-linear response. In particular, we are likely to underestimate the system's response during extreme and rare driving conditions due to uncertainty in the drivers. We demonstrate this phenomenon for extreme space weather and its impact on Earth's magnetosphere, where due to random errors in the measurements of solar wind drivers, there is a non-linear bias in the magnetosphere's response. We propose that the underlying statistical effect (related to the more well-known regression to the mean effect) is generalizable to the other fields that study different systems' responses to driving, like extreme climate studies.