Characterizing precipitation regimes of extreme wet and dry years: Implications for regional to global scale experiments

Background/Question/Methods

Precipitation regimes that characterize extreme wet or dry years differ from average precipitation years in many ways besides amount. Yet, most experiments that manipulate precipitation focus on increasing or decreasing precipitation amount (annual or growing season) by a locally relevant magnitude.   Much less attention is paid to other attributes of precipitation regimes (event size, frequency, distribution, etc.) despite evidence that these may differ substantially between wet and dry years. With recent calls for coordinated network-scale (multi-site) climate change experiments and increased interest in the ecological consequences of climatically extreme events, it is imperative that protocols are designed to incorporate realistic alterations in precipitation regime as well as amount, and that these have broad relevance for multiple biomes. We assessed long-term (100 yr) precipitation records from more than 1200 sites spanning 12 globally important ecoregions (deserts, grasslands, forests, etc.) with mean annual precipitation (MAP) varying from 200-2000 mm. Our goal was to identify key attributes of precipitation regimes that characterize very wet or dry years and distinguish these from average precipitation years.

Results/Conclusions

We found that for sites with > 500 mm MAP, precipitation amount in very wet and dry years consistently deviated by 30-35% from the long-term mean. Below 500 mm MAP, deviations increased asymmetrically - up to 75% for dry years and > 100% for wet years in the driest sites. Precipitation regimes of very wet and dry years were most consistently characterized across all ecoregions by either many (or very few) large rainfall events (defined as daily precipitation events that statistically exceeded the 95^thpercentile of all events in amount).  In other words, the key difference in precipitation regimes between very wet or dry years was (1) the presence of several extremely large precipitation events in wet years and (2) a low number of these in dry years. Less important was mean event size, event number, or the number of dry days between events.  Thus, to best experimentally simulate very wet or dry years with the broadest relevance to most ecoregions, protocols that include additions or removal of large precipitation events, in addition to altering amount, are recommended.