U.S. Floods, Droughts and Global Warming: Another Wardrobe Failure
Two scientists from the U.S. Geological Survey (USGS), Gregory McCabe and David Wolock, recently examined historical (1951-2009) streamflow records from 516 rivers and streams that they considered to be only minimally impacted by human development. They first sorted the data into regional patterns, and then compared the temporal behavior of these patterns to common historical climate indices—such as well-known patterns of atmospheric circulation, sea surface temperatures, or even large-scale warming.
It turns out that there weren’t any relationships between streamflow and the larger atmospheric phenomena. Or at least, so very few that they are hardly worth mentioning.
Here is how McCabe and Wolock describe what they (didn’t) find:
Comparing time series of climate indices…with the time series of mean [stream] flow for the 14 clusters [patterns] indicates weak correlations that are statistically significant for only a few clusters. These results indicate that most of the temporal variability in streamflow in the conterminous U.S. is unpredictable in terms of relations to well-known climate indices. [emphasis added]
In other words, trends and/or variability in larger-scale features of the climate (including rising temperature from global warming) are not very strongly (if at all) related to regional and temporal characteristics of streamflows across the U.S.
And before anyone starts to argue that we have left out the direct (i.e., local) effect of global warming—that warmer air holds more moisture and thus it can rain more frequently and harder—McCabe and Wolock report very few long-term trends that would be indicative of steadily rising moisture levels. Instead, the find the historical records dominated by periods of multidecadal variability. In their own words:
Analyses of the annual mean streamflow time series for the 14 streamflow clusters indicated periods of extended wet and dry periods, but did not indicate any strong monotonic trends. Thus, the mean cluster streamflow time series indicate nearly random variability with some periods of persistence.
McCabe, G. J., and D. M. Wolock, 2014. Spatial and Temporal Patterns in Conterminous United States Streamflow Characteristics. Geophysical Research Letters, doi:10.1002/2014GL061980