Strategies for development of time-varying intensity-duration-frequency curves

Considerable evidence is accumulating that frequency distributions of precipitation amounts have been changing. The greatest agreement among these studies has been in the upper tails of the frequency distributions. Unusually heavy precipitation events are slowly becoming more frequent. This has potentially important implications for design and construction of a broad range of physical infrastructure. The engineering profession typically obtains design standards by consulting published intensity-duration-frequency (IDF) curves. The cost associated with designing for low-probability / high-consequence events is typically quite high. Thus, changes in these standards, especially if extremes become “more extreme,” likewise incur great added cost. For these reasons, design values and the methods that produced them must be acceptable to two different audiences: 1) the scientific community (intellectual rigor, accuracy), and 2) the engineering and planning communities and also the political process (understandability, transparency, practicality, implementability). Traditional practice, based on observational records, has been to assume that climate statistics are stationary. The clear implication of the aforementioned research is that they are not. It also appears that these statistics, and the resulting IDF curves, will slowly evolve with time, and could potentially be thought of as time-dependent. This raises a number of practical questions that relate to the intended interval for applicability. There is no single such interval, but rather a wide range of intervals (often a decade or two, to a century or more). For long enough intervals of application, the statistics may change even within that interval. These and other considerations will be discussed, and a strategy for approaching this issue will be outlined.