Characterizing the responses of local floods to changing climate in three different hydroclimatic regions across the United States

How floods respond to changing climate is of great concern in research community and society, but there is no consistent conclusion about whether floods are as sensitive to changes in temperature as precipitation extremes. In this study, the responses and connections of local floods to temperature and precipitation extremes over three hydroclimatically different regions across the United States were explored. The regions include Northeast Coast (humid coast), Central Plains (dry inland), Southeast Plains (humid subtropic), varying in atmospheric moisture availability. Investigation was made at both hourly and daily timescales, and the influences of watershed imperviousness on flood responses were highlighted by using small watersheds (< 600 km(2)) and categorizing watersheds based on watershed imperviousness. Correlation and co-occurrence probability analysis shows that the connections between floods and precipitation extremes over the studied watersheds are relatively high, with the majority showing a value of 0.6 and 0.5 for the two criteria, respectively. Scaling analysis finds that hourly precipitation extremes generally scale with daily temperature according to the Clausius-Clapeyron (CC) relationship, but daily precipitation extremes reflect smaller scaling rates and even decrease with rising temperature at higher temperatures. Despite floods in the studied watersheds are highly related with precipitation extremes, the scaling of flood-temperature is always lower than that of extreme precipitation-temperature. Increasing connections between floods and precipitation extremes, resulted from higher watershed imperviousness or in summer months, tend to increase the scaling of flood-temperature, even to a close CC relationship. However, the scaling of flood-temperature still cannot surpass that of extreme precipitation-temperature. It is generally concluded that flood would change with rising temperature at a rate smaller than what is expected from the scaling of precipitation extremes to temperature, and precipitation scaling analysis should better use finer-timescale (e.g., hourly) precipitation records.