A global assessment of changes in 20-year, 50-year, and 100-year river flood return periods

Changes in climate, land cover, and human activities are altering the spatial and temporal characteristics of hydrological extremes globally. As a result, many have questioned the validity of using time-invariant probability distributions to estimate the risk of flooding. Here we use a nonstationary approach to assess changes in flooding at the national and global scale. We model annual maximum daily streamflow to evaluate changes in: (a) the magnitudes of the 20-, 50-, and 100-year floods (i.e. flows of a given exceedance probability in each year); and (b) the return periods of the 20-, 50-, and 100-year floods, as assessed in 1970 (i.e. flows of a fixed magnitude). We evaluate thousands of observed stream gauge records obtained from a broad array of national and international archives.

Our analysis suggests that the 20-, 50-, and 100- year flood return periods estimated for conditions in the 1970s have experienced wide spatial redistribution under present-day conditions (thereby indicating increasing/decreasing flood risk). These estimates vary considerably by country and by record length: we find a notable spatial redistribution of flood risk, with some regions witnessing substantial decreases in extreme flood return periods (increased risk of high flows; e.g., UK, Belgium, France, Switzerland), and other countries witnessing increases in extreme return periods (decreased risk of high flows; e.g., Spain, Northeastern Europe). Many countries exhibit a wide range of both increasing and decreasing extreme river flood flows, with some coherent regional patterns (e.g., United States, China, Brazil, Mexico, Australia, South Africa, and Canada). Our methods are transferable to other hazards such as extreme temperature, precipitation, and wind, and indicate that better estimates of local risk for communities require that the probabilities of environmental extremes be updated regularly.