Skip to content

It it’s 2007 Fourth Assessment Report (AR4), the Intergovernmental Panel on Climate Change (IPCC) investigated the character of the changes in precipitation that have been observed across the world over the past half-century or so. In doing so, they concluded:

Observed changes in intense precipitation (with geographically varying thresholds between the 90th and 99.9th percentile of daily precipitation events) for more than one half of the global land area indicate an increasing probability of intense precipitation events beyond that expected from changes in the mean for many extratropical regions (Groisman et al., 2005; Figure 3.39, bottom panel). This finding supports the disproportionate changes in the precipitation extremes described in the majority of regional studies above, in particular for the mid-latitudes since about 1950.

Here’s the graphic mentioned in the quote above:

Figure 1. Bottom panel of IPCC AR4 Figure 3.39 showing “regions where disproportionate changes in heavy and very heavy precipitation during the past decades were documented as either an increase (+) or decrease (–) compared to the change in the annual and/or seasonal precipitation” (from IPCC AR4, 2007).

The problem is, is that the methodology that the IPCC relied upon was insufficient to base such a conclusion.

Over those regions where a more appropriate methodology has been employed, contrary to IPCC proclamations, the changes in intense precipitation events have proven to be very much proportionate to the observed changes in total annual precipitation. And even though the IPCC knew that this was the case for at least one of these major land areas (the United States), they nevertheless chose to forward the mischaracterization.

Chalk up another addition to the list of IPCC errors.

Good and Bad Methodologies

In virtually all assessments of precipitation changes, the first thing that is done, is to determine the average amount (or number) of daily precipitation (events) that fall within a predefined set of “bins.” For instance, how much precipitation falls on average in daily events between one-half and one inch, or between one and two inches, or greater than two inches. The bin thresholds are often defined by percentiles. For instance the amount of rain that falls between the 90th and 95th percentile of all rain events, or between the 95th and 99th percentile, etc. These thresholds remain fixed throughout the analysis. Then, the amount of rain that falls within each of these bins during each year over the period of analysis is examined to determine if there have been any overall changes.

Based on the results of analyses using this fixed bin approach, it is often concluded that the amount of rain falling in “heavy,” or “intense,” or “extreme” events is increasing at a rate that is disproportionate to (increasing faster than) the increase in total annual precipitation. And such conclusions base the IPCC assessment of the situation.

However, the shape of the natural distribution of daily rainfall events means that a fixed bin methodology will find disproportionate changes even when the underlying changes are perfectly proportionate to each other and to the total annual change.

We explained all of this in a paper we published (Michaels et al., 2004) back in 2004 in the International Journal of Climatology (after a long and contentious review process at Geophysical Research Letters in which we were ultimately rejected), and cautioned that fixed bin methodologies “cannot differentiate between proportionate and disproportionate precipitation changes in all cases.” And that “[t]o assess the proportionality of observed changes accurately, a different analysis technique must be employed.”

Figure 2 (taken from our paper) shows why this is the case.

Figure 2. (a) Simulated changes in the percentage change of precipitation within 20 five-percentile bins from a 10% increase in the number of events; (b) same as in (a), except for a 10% increase in the intensity of each event; (c) same as is (a), except for a combination of more events and greater intensity events (from Michaels et al., 2004).

We first generated some synthetic data drawn randomly from a “gamma” distribution—a type of distribution that well describes the character of precipitation and which is skewed such more than half of the annual rainfall total is delivered from a relatively few heavy rain events. We used this dataset to define the thresholds for our fixed bins. We then simulated three potential ways in which total annual precipitation could increase by 10% with the increase being perfectly proportionately distributed across all bins—the number of precipitation events was increased by 10%, the intensity of each precipitation event was increased by 10%, and a combination of the two. We then applied the standard methodology (which underlies the IPCC conclusions) to assess how our proportionate changes would be manifest. What we found was that using the standard fixed-bin approach, it appears as if the 10% increase in precipitation is disproportionately dominated by the heavy precipitation events (those lying towards the right-hand side of the figures), even though we know a priori (because we defined it to be the case) that the changes were perfectly proportionate across all bin sizes. Clearly, as reflected in our conclusions quoted above, fixed-bin approaches can give very misleading results.

The question then is do they?

Proportionate Changes

In our paper, we forward a more appropriate technique for assessing changes in precipitation—instead of using fixed bins, we proposed examining trends through ranked precipitation events (that is, the trend of the amount of precipitation that falls on the wettest day of each year, the second wettest day of each year, … the tenth wettest day of each year). We could then compare the changes in the amount of precipitation on each ranked days with the changes in a total annual precipitation.

What we found was that while the amount of precipitation falling on each of the top-10 wettest days of the year had increased in the U.S. from 1910-2001, the increase on those days was, by and large, in perfect proportion to the increase in the total annual rainfall amount.

Figure 3 (taken from our paper) shows what we found for the trend in the amount of precipitation on the wettest day of each year averaged across the country. The top panel show that the absolute amount of precipitation on the wettest day of the year is increasing—that is, the wettest day of the year is getting even wetter. The bottom panel shows the percentage of the annual precipitation total that falls on the wettest day of the year. No change at all. In other words, the increase is total annual rainfall is keeping up with the increase in the rainfall on the wettest day of the year, i.e. the increase has been proportionate.

Figure 3. (a) Time series of the wettest day of the year averaged across the contiguous United States from 1910-2001. (b) Time series of the percentage of total annual precipitation contributed by the wettest day. (source: Michaels et al., 2004)

Obviously, we concluded that labeling the observed change in heavy precipitation events in the U.S. “disproportionate” was extremely misleading.

Nevertheless (and despite our complaints; see comment 3-838 in the public comments submitted to the IPCC) the IPCC proceeded to do so anyway. This is evidenced by the grouping of plus signs (“+”) over the U.S. in out Figure 1 above.

New Results

When we did our work back in 2004, we were only able to examine trends in U.S. precipitation from a fairly limited amount of available data. But in the intervening time, a much greater volume of data has become digitally available, not only for the U.S., but also internationally as well. And now, a team of scientists has performed an analysis similar to ours for China.

And they found for China precipitation the same thing we did for U.S. precipitation—that is, the change in extreme precipitation is generally in line with changes in total precipitation, despite being mischaracterized in other studies as being “disproportionate.”

Here is how Binhui Liu of the the Northeast Forestry University, in Harbin, China and colleagues summarize their work:

Previous studies have suggested that extreme precipitation events accounted for a disproportionate share of the nearly 2% increase in precipitation in China over the period of 1960–2000. Michaels et al. challenged a similar finding in the USA, arguing that fixed-bin methods for analysing extreme events obscure underlying precipitation patterns, and proposing a method that focusses on trends of the 10 wettest days of the year. Applying this method to China, we find that trends of precipitation on the 10 wettest days are generally proportional to changes in annual total precipitation.

Another strike for the IPCC as they have a big “plus” sign over China (see Figure 1) where one does not belong.

We can only wonder what would happen if the appropriate methodology were to be applied to the other regions of the world where the IPCC has deemed the changes in extreme precipitation “disproportionate” to changes in the mean. After all, in the two places that the appropriate methodology was applied, the IPCC’s assessment was found to be inaccurate in both instances.

This is not to say that there have not been increases in extreme precipitation amounts in the U.S or in China. In fact, generally there have been. And documenting these changes is important as extreme precipitation may have influences on flood severity and frequency, agriculture, river navigation, power generation, and a range of other considerations.

However, using rather provocative language to describe them is both inappropriate and misleading. More and more evidence is turning up that the only thing “disproportionate” about the changes in precipitation has been amount of spin that has been put on an expected and normal climate behavior.

When it rains more, more rain falls in heavy events. While this information is good to know, it is not an unnatural sign of a human influence on the climate—it is just the way it is. For the vast majority of the world and its increasing water use demands, more precipitation is a good thing. If human activities are contributing to the increase, we at World Climate Report kind of think this sounds generally like a good thing. Silly us.


Intergovernmental Panel on Climate Change, 2007. Climate Change 2007, The Physical Science Basis, Working Group I Contribution to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Solomon, S.D., et al., (eds), Cambridge University Press, Cambridge, UK and New York, 996 pp.

Liu, B., et al., 2011. Observed changes in precipitation on the wettest days of the year in China, 1960-2000. International Journal of Climatology31, 487-503.

Michaels P.J., Knappenberger, P.C., Frauenfeld, O.W., and R.E. Davis RE., 2004. Trends in precipitation on the wettest days of the year across the contiguous USA. International Journal of Climatology24,1873–1882.

World Climate Report, 11 October 2011