Learn how plants respond to higher atmospheric CO2 concentrations

How does rising atmospheric CO2 affect marine organisms?

Click to locate material archived on our website by topic


Extreme Weather Events: Are they Influenced by Rising Atmospheric CO2?

3.1.2. Natural Flood Variability Seen from Long-term, Centennial-scale Studies


Beyond short term analyses of only a few decades, a number of studies have examined flooding over centennial to millennial time scales. These studies, which comprise those reviewed in this section, allow the comparison of flood events that occurred prior to the modern buildup of anthropogenic CO2 in the air with those that occurred after it. These types of analyses reveal great detail about the breadth and depth of natural variability and are of great value in investigating the potential influence of rising CO2 on floods.

During the 1990s, broad areas of the US Northern Great Plains experienced notable lake highstands, including Waubay Lake, which rose by 5.7 meters and more than doubled in area from 1993 to 1999, severely flooding roads, farms and towns, and prompting the Federal Emergency Management Agency to declare the region a disaster area on 1 June 1998. Shapley et al. (2005) set out to determine the historical context of that 1990s lake-level rise by developing a 1,000-year hydroclimate reconstruction from local bur oak (Quercus macrocarpa) tree-ring records and lake sediment cores from the Waubay Lake complex located in eastern South Dakota. In doing so, the researchers found that "prior to AD 1800, both lake highstands and droughts tended towards greater persistence than during the past two centuries," such that "neither generally low lake levels occurring since European settlement (but before the recent flooding) nor the post-1930s pattern of steadily increasing water availability and favorableness for tree growth are typical of the long-term record." In this particular part of the world, therefore, it is clear that longer-lasting floods and droughts of equal or greater magnitude than those of modern times occurred repeatedly prior to 1800 and independent of atmospheric CO2.

A bit further to the north, significant flooding of the Red River of the North occurred in 1997, which devastated Grand Forks, North Dakota, as well as parts of Canada. However, as Haque (2000) reports, although this particular flood was indeed the largest experienced by the Red River over the past century, it was not the largest to occur in historic times. In 1852, for example, there was a slightly larger Red River flood; and in 1826 there was a much larger flood that was nearly 40% greater than the flood of 1997. And the temperature of the globe, it should be noted, was much colder at the times of these earlier catastrophic floods than it was in 1997, indicating the strength of the 1997 flood cannot be attributed to the degree of warmth experienced that year or throughout the preceding decade.

In a study designed to determine the environmental origins of extreme flooding events throughout the southwestern United States, Ely (1997) wrote that "paleoflood records from nineteen rivers in Arizona and southern Utah, including over 150 radiocarbon dates and evidence of over 250 flood deposits, were combined to identify regional variations in the frequency of extreme floods," which information "was then compared with paleoclimatic data to determine how the temporal and spatial patterns in the occurrence of floods reflect the prevailing climate." The results of this comparison indicated that "long-term variations in the frequency of extreme floods over the Holocene are related to changes in the climate and prevailing large-scale atmospheric circulation patterns that affect the conditions conducive to extreme flood-generating storms in each region," which changes, in Ely's view, "are very plausibly related to global-scale changes in the climate system."

With respect to the Colorado River watershed, for example, which integrates a large portion of the interior western United States, she writes that "the largest floods tend to be from spring snowmelt after winters of heavy snow accumulation in the mountains of Utah, western Colorado, and northern New Mexico," such as occurred with the "cluster of floods from 5 to 3.6 ka," which occurred in conjunction with "glacial advances in mountain ranges throughout the western United States" during the "cool, wet period immediately following the warm mid-Holocene."

The frequency of extreme floods also increased during the early and middle portions of the first millennium AD, many of which coincided "with glacial advances and cool, moist conditions both in the western U.S. and globally." Then came a "sharp drop in the frequency of large floods in the southwest from AD 1100-1300," which corresponded, in her words, "to the widespread Medieval Warm Period, which was first noted in European historical records." With the advent of the Little Ice Age, however, there was another "substantial jump in the number of floods in the southwestern U.S.," which was "associated with a switch to glacial advances, high lake levels, and cooler, wetter conditions." And in distilling her findings down to a single succinct statement, and speaking specifically of the southwestern United States, Ely states that "global warm periods, such as the Medieval Warm Period, are times of dramatic decreases in the number of high-magnitude floods in this region." This conclusion, of course, directly contradicts the climate alarmist hypothesis that warmer temperatures results in increased extreme flood events.

In another study encompassing the entire continental United States, Fye et al. (2003) developed multi-century reconstructions of summer (June-August) Palmer Drought Severity Index from annual proxies of moisture status provided by 426 climatically-sensitive tree-ring chronologies. This exercise indicated that the greatest 20th-century wetness anomaly across the United States was the 13-year pluvial that occurred in the early part of the century, when it was considerably colder than it is now. In addition, Fye et al.'s analysis revealed the existence of a 16-year pluvial from 1825 to 1840 and a prolonged 21-year wet period from 1602 to 1622, both of which anomalies occurred during the Little Ice Age, when, of course, it was colder still.

Moving to analyses from other countries, on the 8th and 9th of September 2002, extreme flooding of the Gardon River in southern France occurred as half an average year's rainfall was received in approximately twenty hours, which flooding claimed the lives of a number of people and caused much damage to towns and villages situated adjacent to its channel. The event garnered much press coverage; and, in the words of Sheffer et al. (2003a), "this flood is now considered by the media and professionals to be 'the largest flood on record'," which record extended all the way back to 1890.

Coincidently, Sheffer et al. were in the midst of a study of prior floods of the Gardon River when the "big one" hit; and they had data spanning a much longer time period against which to compare its magnitude. Based on that data as presented in their paper, they reported "the extraordinary flood of September 2002 was not the largest by any means," noting "similar, and even larger floods have occurred several times in the recent past," with three of the five greatest floods they had identified to that point in time occurring over the period AD 1400-1800 during the Little Ice Age. Commenting on these facts, Sheffer et al. state that "using a longer time scale than human collective memory, paleoflood studies can put in perspective the occurrences of the extreme floods that hit Europe and other parts of the world during the summer of 2002." And that perspective clearly shows that even greater floods occurred repeatedly during the Little Ice Age, which was the coldest period of the current interglacial and which obviously had nothing to do with atmospheric CO2.

Working in the same region five years later, Sheffer et al. (2008) analyzed geomorphic, sedimentologic and hydrologic data associated with both historical and late Holocene floods from two caves and two alcoves of a 1600-meter-long stretch of the Gardon River, which analysis they hoped would provide a longer and better-defined perspective on the subject. And so it did, as they discovered "at least five floods of a larger magnitude than the 2002 flood occurred over the last 500 years," all of which took place, as they describe it, "during the Little Ice Age." In addition, they note that "the Little Ice Age has been related to increased flood frequency in France (Guilbert, 1994; Coeur, 2003; Sheffer, 2003; Sheffer et al., 2003a,b; Sheffer, 2005), and in Spain (Benito et al., 1996; Barriendos and Martin Vide, 1998; Benito et al., 2003; Thorndycraft and Benito, 2006a,b)."

Introducing their study of the subject, Wilhelm et al. (2012) write "mountain-river floods triggered by extreme precipitation events can cause substantial human and economic losses (Gaume et al., 2009)," and they say "global warming is expected to lead to an increase in the frequency and/or intensity of such events (IPCC, 2007), especially in the Mediterranean region (Giorgi and Lionello, 2008)." However, they caution that "reconstructions of geological records of intense events are an essential tool for extending documentary records beyond existing observational data and thereby building a better understanding of how local and regional flood hazard patterns evolve in response to changes in climate."

In an effort to obtain this "better understanding," Wilhelm et al. analyzed the sediments of Lake Allos, a 1-km-long by 700-m-wide high-altitude lake in the French Alps (4414'N, 642'35"E), by means of both seismic survey and lake-bed coring, carrying out numerous grain size, geochemical and pollen analyses of the sediment cores they obtained in conjunction with a temporal context derived using several radionuclide dating techniques. In doing so, the thirteen researchers, all hailing from France, report their investigations revealed the presence of some 160 graded sediment layers over the last 1,400 years; and they indicate comparisons of the most recent of these layers with records of historic floods suggest the sediment layers are indeed representative of significant floods that were "the result of intense meso-scale precipitation events." Of special interest to the discussion at hand is their finding of "a low flood frequency during the Medieval Warm Period and more frequent and more intense events during the Little Ice Age," which meshes nicely with the results of an analysis of a Spanish lake sediment archive that allowed Moreno et al. (2008) to infer "intense precipitation events occurred more frequently during the Little Ice Age than they did during the Medieval Warm Period."

Wilhelm et al. additionally state that "the Medieval Warm Period was marked by very low hydrological activity in large rivers such as the Rhone (Arnaud et al., 2005; Debret et al., 2010), the Moyenne Durance (Miramont et al., 1998), and the Tagus (Benito et al., 2003), and in mountain streams such as the Taravilla lake inlet (Moreno et al., 2008)." But of the Little Ice Age, they say "research has shown higher flood activity in large rivers in southern Europe, notably in France (Miramont et al., 1998; Arnaud et al., 2005; Debret et al., 2010), Italy (Belotti et al., 2004; Giraudi, 2005) and Spain (Benito et al., 2003), and in smaller catchments (e.g., in Spain, Moreno et al., 2008)."

In concluding their report, Wilhelm et al. say their study shows "sediment sequences from high altitude lakes can provide reliable records of flood-frequency and intensity-patterns related to extreme precipitation events," closing with the warning that "such information is required to determine the possible impact of the current phase of global warming." And when this warning is heeded, it is clearly seen that the climate-model-inspired claim that global warming will lead to "an increase in the frequency and/or intensity of such events"-would appear to be just the opposite of what is suggested by Wilhelm et al.'s real-world study and the real-world studies of the other scientists they cite.

Glur et al. (2103) developed "a multi-archive Alpine flood reconstruction based on ten lacustrine sediment records, covering the past 2,500 years" for the European Alps. In discussing their findings the eight researchers report "flood activity was generally enhanced during the Little Ice Age (1430-1850 C.E.; LIA) compared to the Medieval Climate Anomaly (950-1250 C.E.; MCA)." And they say "this result is confirmed by other studies documenting an increased (decreased) flood activity during the LIA (MCA) in the Alps," citing the studies of Schmocker-Fackel and Naef (2010), Czymzik et al. (2010), Wilhelm et al. (2012) and Swierczynski et al. (2012). Thus, for the European Alps, there would appear to be good reason to conclude that any further warming of the globe would not lead to flood-induced "increased threats to settlements, infrastructure, and human lives," for real-world data suggest that it is cooling that leads to such consequences in that part of the world.

Focusing on the region of southwest Germany, Burger et al. (2007) reviewed what is known about flooding in this region over the past three centuries. According to the six scientists, the extreme flood of the Neckar River (southwest Germany) in October 1824 was "the largest flood during the last 300 years in most parts of the Neckar catchment." In fact, they say "it was the highest flood ever recorded in most parts of the Neckar catchment and also affected the Upper Rhine, the Mosel and Saar." In addition, they report that the historical floods of 1845 and 1882 "were among the most extreme floods in the Rhine catchment in the 19th century," which they describe as truly "catastrophic events." And speaking of the flood of 1845, they say it "showed a particular impact in the Middle and Lower Rhine and in this region it was higher than the flood of 1824." Finally, the year 1882 actually saw two extreme floods, one at the end of November and one at the end of December. Of the first one, Burger et al. say that "in Koblenz, where the Mosel flows into the Rhine, the flood of November 1882 was the fourth-highest of the recorded floods, after 1784, 1651 and 1920," with the much-hyped late-20th-century floods of 1993, 1995, 1998 and 2002 not even meriting a mention.

"Starting from historical document sources, early instrumental data (basically, rainfall and surface pressure) and the most recent meteorological information," as they describe it, Llasat et al. (2005) analyzed "the temporal evolution of floods in NE Spain since the 14th century," focusing particularly on the river Segre in Lleida, the river Llobregat in El Prat, and the river Ter in Girona. This work indicated there was "an increase of flood events for the periods 1580-1620, 1760-1800 and 1830-1870," and they report that "these periods are coherent with chronologies of maximum advance in several alpine glaciers." In addition, it can be calculated from their tabulated data that, for the aggregate of the three river basins noted above, the mean number of what Llasat et al. call catastrophic floods per century for the 14th through 19th centuries was 3.55 ± 0.22, while the corresponding number for the 20th century was only 1.33 ± 0.33.

In concluding their paper, the four Spanish researchers say "we may assert that, having analyzed responses inherent to the Little Ice Age and due to the low occurrence of frequent flood events or events of exceptional magnitude in the 20th century, the latter did not present an excessively problematic scenario." However, having introduced their paper with descriptions of the devastating effects of the September 1962 flash flood in Catalonia (over 800 deaths), the August 1996 flash flood in the Spanish Pyrenees (87 deaths), as well as the floods of September 1992 that produced much loss of life and material damage in France and Italy, they hastened to add that the more recent "damage suffered and a perception of increasing vulnerability is something very much alive in public opinion and in economic balance sheets."

Shifting to the area of southeast Spain, Benito et al. (2010) reconstructed flood frequencies of the Upper Guadalentin River using "geomorphological evidence, combined with one-dimensional hydraulic modeling and supported by records from documentary sources at Lorca in the lower Guadalentin catchment." The combined palaeoflood and documentary records indicate that past floods were clustered during particular time periods: AD 950-1200 (10), AD 1648-1672 (10), AD 1769-1802 (9), AD 1830-1840 (6), and AD 1877-1900 (10), where the first time interval coincides with the Medieval Warm Period and the latter four time intervals all fall within the confines of the Little Ice Age; and by calculating mean rates of flood occurrence over each of the five intervals, a value of 0.40 floods per decade during the Medieval Warm Period and an average value of 4.31 floods per decade over the four parts of the Little Ice Age can be determined, which latter value is more than ten times greater than the mean flood frequency experienced during the Medieval Warm Period.

Introducing their study of the subject, Stewart et al. (2011) note that "regional climate models project that future climate warming in Central Europe will bring more intense summer-autumn heavy precipitation and floods as the atmospheric concentration of water vapor increases and cyclones intensify," citing the studies of Arnell and Liu (2001), Christensen and Christensen (2003) and Kundzewicz et al. (2005). In an exercise designed to assess the reasonableness of these projections, Stewart et al. derived "a complete record of paleofloods, regional glacier length changes (and associated climate phases) and regional glacier advances and retreats (and associated climate transitions) from the varved sediments of Lake Silvaplana (ca. 1450 BC-AD 420; Upper Engadine, Switzerland)," while indicating that "these records provide insight into the behavior of floods (i.e. frequency) under a wide range of climate conditions."

Based on their analysis, the five researchers report there was "an increase in the frequency of paleofloods during cool and/or wet climates and windows of cooler June-July-August temperatures" and that the frequency of flooding "was reduced during warm and/or dry climates." And reiterating the fact that "the findings of this study suggest that the frequency of extreme summer-autumn precipitation events (i.e. flood events) and the associated atmospheric pattern in the Eastern Swiss Alps was not enhanced during warmer (or drier) periods," Stewart et al. acknowledge that "evidence could not be found that summer-autumn floods would increase in the Eastern Swiss Alps in a warmer climate of the 21st century," in contrast to the projections of the regional climate models that have suggested otherwise.

Mudelsee et al. (2004) prefaced their work by writing "extreme river floods have had devastating effects in central Europe in recent years," citing as examples the Elbe flood of August 2002, which caused 36 deaths and inflicted damages totaling over 15 billion U.S. dollars, and the Oder flood of July 1997, which caused 114 deaths and inflicted approximately 5 billion dollars in damages. And they noted that concern had been expressed in this regard "in the Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change," wherein it was stated that "current anthropogenic changes in atmospheric composition will add to this risk."

Unconvinced about this contention, the four researchers reevaluated the quality of data and methods of reconstruction that had previously produced flood histories of the middle parts of the Elbe and Oder rivers back to AD 1021 and 1269, respectively; and in doing so, they found, for both the Elbe and Oder rivers, "no significant trends in summer flood risk in the twentieth century," but "significant downward trends in winter flood risk," which latter phenomenon-described by them as "a reduced winter flood risk during the instrumental period"-they specifically described as "a response to regional warming." Thus, their study provided no support whatsoever for the IPCC "concern" that CO2-induced warming would add to the risk of river flooding in Europe. If anything, their findings suggested just the opposite.

Writing as background for their work, Buntgen et al. (2011) correctly indicate that instrumental station measurements, which systematically cover only the last 100-150 years, "hinder any proper assessment of the statistical likelihood of return period, duration and magnitude of climatic extremes," stating that "a palaeoclimatic perspective is therefore indispensable to place modern trends and events in a pre-industrial context (Battipaglia et al., 2010), to disentangle effects of human greenhouse gas emission from natural forcing and internal oscillation (Hegerl et al., 2011), and to constrain climate model simulations and feedbacks of the global carbon cycle back in time (Frank et al., 2010)." In an effort to satisfy these requirements and help facilitate the accomplishment of the associated goals, Buntgen et al. "introduce and analyze 11,873 annually resolved and absolutely dated ring width measurement series from living and historical fir (Abies alba Mill.) trees sampled across France, Switzerland, Germany and the Czech Republic, which continuously span the AD 962-2007 period," and which "allow Central European hydroclimatic springtime extremes of the industrial era to be placed against a 1,000 year-long backdrop of natural variations."

In the words of the nine researchers, their data revealed "a fairly uniform distribution of hydroclimatic extremes throughout the Medieval Climate Anomaly, Little Ice Age and Recent Global Warming." Such finding, as stated by the authors, "may question the common belief that frequency and severity of such events closely relates to climate mean states," which conclusion represents a rebuke of the claim that global warming will lead to more frequent and severe floods and droughts.

Lindstrom and Bergstrom (2004) analyzed runoff and flood data from more than 60 discharge stations scattered throughout Sweden, some of which provided information stretching as far back in time as the early to mid-1800s, when Sweden and the world were still experiencing the cold of the Little Ice Age. This analysis led them to discover that the last 20 years of the past century were indeed unusually wet, with a runoff anomaly of +8% compared with the century average. But they also found "the runoff in the 1920s was comparable to that of the two latest decades," and "the few observation series available from the 1800s show that the runoff was even higher than recently." In addition, they determined "flood peaks in old data [were] probably underestimated," which "makes it difficult to conclude that there has really been a significant increase in average flood levels." Also, they say "no increased frequency of floods with a return period of 10 years or more, could be determined."

With respect to the generality of their findings, Lindstrom and Bergstrom concluded that conditions in Sweden "are consistent with results reported from nearby countries: e.g. Forland et al. (2000), Bering Ovesen et al. (2000), Klavins et al. (2002) and Hyvarinen (2003)," and that, "in general, it has been difficult to show any convincing evidence of an increasing magnitude of floods (e.g. Roald, 1999) in the near region, as is the case in other parts of the world (e.g. Robson et al., 1998; Lins and Slack, 1999; Douglas et al., 2000; McCabe and Wolock, 2002; Zhang et al., 2001)."

In Asia, Davi et al. (2006) developed a reconstruction of streamflow that extended from 1637 to 1997, based on absolutely dated tree-ring-width chronologies from five sampling sites in west-central Mongolia, all of which sites were in or near the Selenge River basin, the largest river in Mongolia. Of the ten wettest five-year periods, only two occurred during the 20th century (1990-1994 and 1917-1921, the second and eighth wettest of the ten extreme periods, respectively), once again indicative of a propensity for less flooding during the warmest portion of the record.

In a study of the Yangtze Delta, Zhang et al. (2007) developed flood and drought histories of the past thousand years "from local chronicles, old and very comprehensive encyclopedia, historic agricultural registers, and official weather reports," after which "continuous wavelet transform was applied to detect the periodicity and variability of the flood/drought series" and, finally, the results of the entire set of operations were compared with 1000-year temperature histories of northeastern Tibet and southern Tibet. This work revealed "colder mean temperature in the Tibetan Plateau usually resulted in higher probability of flood events in the Yangtze Delta region," and the authors say that "during AD 1400-1700 [the coldest portion of their record, corresponding to much of the Little Ice Age], the proxy indicators showing the annual temperature experienced larger variability (larger standard deviation), and this time interval exactly corresponds to the time when the higher and significant wavelet variance occurred." In contrast, they report that "during AD 1000-1400 [the warmest portion of their record, corresponding to much of the Medieval Warm Period], relatively stable climatic changes reconstructed from proxy indicators in Tibet correspond to lower wavelet variance of flood/drought series in the Yangtze Delta region."

In another study focusing on the Yangtze Delta, Zhang et al. (2009) utilized wavelet analysis on the decadal locust abundance data of Ma (1958) for the AD 950s-1950s, the decadal Yangtze Delta flood and drought frequency data of Jiang et al. (2005) for the AD 1000s-1950s, and the decadal mean temperature records of Yang et al. (2002) for the AD 950s-1950s, "to shed new light on the causal relationships between locust abundance, floods, droughts and temperature in ancient China." In doing so, the international team of Chinese, French, German and Norwegian researchers found that coolings of 160-170-year intervals dominated climatic variability in China over the past millennium, and that these cooling periods promoted locust plagues by enhancing temperature-associated drought/flood events. As a result, the six scientists say that "global warming might not only imply reduced locust plague[s], but also reduced risk of droughts and floods for entire China," noting that these findings "challenge the popular view that global warming necessarily accelerates natural and biological disasters such as drought/flood events and outbreaks of pest insects," as promulgated by the most recent report of the Intergovernmental Panel on Climate Change. Indeed, they say their results are an example of "benign effects of global warming on the regional risk of natural disasters."

Taken together, the studies referenced above clearly demonstrate a lack of evidence for the hypothesis that CO2-induced global warming is increasing the frequency and magnitude of flood events. If anything, it suggests flooding tends to be reduced and less severe when the planet experienced warmer, as opposed to colder, temperatures.

Back to the Table of Contents