The notion that human-induced climate change will make for more extreme weather has become writ large on the public consciousness. It makes for good headlines, so surely it must be true. Well, new analyses suggest it might be false (at least for the United States).
Ken Kunkel of the Illinois State Water Survey recently presented some new results (updating some of his earlier work) and a summary of some recent papers on extreme weather in the continental United States at the Climate Specialty Group’s plenary session at the Association of American Geographers annual conference in Chicago. His analysis shed some new light on long-term extremes in heavy precipitation and heat and cold waves based on the availability of some new data.
For over a century, a national network of “weather nerds” (for lack of a better term) have monitored backyard weather stations where they kept track of daily maximum and minimum temperature and precipitation using standardized instruments and measurement techniques. Called the U.S. Cooperative Observer Network (co-op for short), these data, which were submitted monthly for many decades on paper logs, were often used to fill in gaps from the more comprehensive observations taken by trained weather service employees at far fewer locations. But the utility of the co-op records to climate analysis was limited by their cumbersome, paper format. However, recently the interest in climate change spurred the government to digitize these paper records, thus adding many new stations to the existing network. With the addition of the co-op data, the number of stations from roughly 1890 to 1947 doubled or tripled relative to the previous baseline.
These updated records shed new light on the behavior of U.S. extremes. For example, Figure 1 shows the likelihood that the standard 1-day precipitation event was exceeded, in a given year, using 1-year, 5- year, and 20-year baselines. Positive values of the Extreme Precipitation Index (or EPI) indicate above average occurrence of extreme events, while negative values depict less frequent extremes. The data since 1950 shows a clear positive trend that seems to be getting more extreme later in the record, with the last few years showing the greatest extremes. This fits very nicely with common journalistic sentiments that our climate is obviously in never-been-to-before territory. But inclusion of the pre-1950 data paints quite a different picture. Not only did the frequency of extremes vary markedly in the early 20th century days of very low greenhouse gas levels, but the frequency of extreme events in the late 1890s was at least comparable to that in our current climate. Kunkel did some statistical tests demonstrating that the most recent period (1983-2004) was not statistically different from the earliest period (1895-1916) for many combinations of event severity and return period, although a few were significantly different. The bottom line here? The assumption that U.S. rainfall is clearly getting more extreme because of global warming is hardly obvious based on the new and improved record.
Figure 1. The observed exceedances of 1-year, 5-year, and 20-year recurrence intervals of 1-day precipitation across co-op observing stations in the United States. The more positive the EPI anomaly, the greater the incidences of extreme daily precipitation.
Kunkel also looked at heat waves and cold waves, defined by a 4-day event with a 5-year recurrence. The heat wave record (Figure 2) is dominated by the huge spike during the 1930s “Dust Bowl” era. In fact, the recent period is hardly noticeable in the longer-term context, even though the number of heat waves has increased recently compared to the cool summers of the 1960s and 1970s. For cold waves (Figure 3), it’s hard to know what kind of signal we should even expect. Whereas most climatologists argue sensibly that we should expect fewer cold waves as our winter air masses warm, some argue, particularly during the few major cold outbreaks that occur every year, that global warming will produce more extreme heat AND cold. Well, this figure supports neither case. If more cold waves are harbingers of global warming, then the peaks that dominated that 1980s have completely disappeared. And if we should expect fewer cold outbreaks, then how does one account for all the cold air outbreaks 1980s when the atmosphere had plenty of greenhouse gases? The cold wave record shows some interesting long-term variability but no obvious trend.
Figure 2. The observed frequency of exceedances of the 5-year recurrence interval of 4-day heat waves across U.S. co-op stations.
Figure 3. The observed frequency of exceedances of the 5-year recurrence interval of 4-day cold waves across U.S. co-op stations.
The new long-term U.S. precipitation records augmented by the addition of co-op data provide a very interesting picture of how variable U.S. weather has been over the past century. While it’s true that in some cases and for some variables you can make a case that we’re currently experiencing some unusual weather, that case does not apply to all variables and time periods. In short, it’s hard to separate the extremes we’re now experiencing from the high variability seen at other times in the late 1890s and 1900s.
You might argue that not too much should emphasis should be placed on the United States weather records. The Lower 48 account for only 2 percent of Earth’s surface. But this is a cautionary tale about data. The U.S. has the most dense long-term weather records of any similar-sized place on the planet. Note how the addition of early observations can completely change one’s interpretation of the trend (as in Figure 1). If we are faced with such uncertainty with the world’s best data set, how much confidence can we really place in our interpretations of the very sparse records from Africa, Asia, and South America, not to mention the paucity of records from the world’s oceans? As with all issues of global climate change, the devil’s in the details, but details about weather records hardly make for blockbuster headlines.