On an old film strip of rainfall, only the brightest frames are visible, and between them lies noise. A meteorologist assessing flood risk long looked only at such frames, that is, at annual rainfall maxima. Everything else, thousands of daily observations, remained off-screen. The new method develops the entire film, and details that were previously invisible come into view.
The first step is like tuning each microphone in a studio before recording an orchestra. For each of California's 35 weather stations, a curve is fitted individually to describe the tail of the rainfall distribution. In the tail live rare but destructive downpours. What matters here is not just the average rainfall, but how quickly the probability of extreme values decays. The fitting uses daily threshold exceedances rather than annual peaks, so the curve is supported by hundreds of points instead of 55.
The second step brings the connections between stations into focus. Rain in Napa and rain in Winters often arrive together because they are born from the same atmospheric front. To measure this connection, the method no longer uses daily records but annual maxima at pairs of stations. Pairwise comparison is like listening to duets: if two instruments always sound loud at the same time, there is a common cause between them. This is how the synchrony of extremes at different points on the map is assessed.
The result is visible on the risk map. The traditional approach, relying only on annual maxima, found a significant El Niño influence at 14 stations. The new method detected this effect at 22 stations. Confidence intervals for joint 50-year return levels narrowed by 27–51%. This means that an insurer or a city planner gets not a blurred patch but a sharp line marking where the zone of acceptable risk ends.
The old film is finally fully developed. Daily records, once dismissed as noise, turned out to be signal. The method does not require complex multivariate models and fits easily into the familiar toolkit of hydrologists. The next frame can already be shot with a more sensitive film, and even the rarest downpours will no longer come as a surprise.
The joint 50-year return level for the station pair Napa and Winters at a Southern Oscillation Index value of -1 was about 10–15 cm of precipitation per day, depending on the estimation method.