Quantifying Observed Changes in Global Precipitation Variability and Extremes

Anthropogenic climate change is driving shifts in precipitation around the globe. Subsequent fluctuations in extreme precipitation can drive substantial impacts through flooding and drought, resulting in enhanced dangers to human well-being and extensive hazards to infrastructure that was largely designed and built under an assumption of a stationary climate. In addition to the effects of changing precipitation extremes, changes in precipitation variability can have profound implications. For example, recent research found that precipitation variability is a constraint on pastoral vegetation and can drive adverse birth outcomes in Amazonia. We use historical precipitation observations from the Global Historical Climatology Network (GHCN-Daily) to examine station-level changes in precipitation for more than 5800 stations with 50-plus years of qualifying data. Trends in precipitation variability, extremes, frequency, and dry spell length, are investigated at annual, seasonal, and daily time scales using non-parametric methods. Roughly 40% of all qualifying stations observed statistically significant changes in annual mean precipitation, precipitation frequency, and annual mean dry spell length, while about 20% report significant changes in annual precipitation variability and 99th percentile precipitation events. However, these changes are not spatially uniform nor are they consistent across metrics. Paradoxically, while annual precipitation generally increases around the globe, changes in variability are mixed. For instance, stations in Europe and India both primarily recorded increases in precipitation, but decreases in precipitation variability. Precipitation whiplash events, a sub-20th percentile precipitation year followed by a super-80th percentile precipitation year, are also explored. Ultimately, our analysis identifies global and regional trends in precipitation variability and extremes and provides a point of comparison for historical model-observation comparison.