Kevin Trenberth’s Climate Attribution Studies versus Useful Science - Part 1 Hurricane Sandy

Useful climate science helps humanity adapt to natural weather patterns and plan for future extremes. Minimizing risk would be a wise course of action, but too often humans have ignored nature’s indicators of high flood risk and continue to build on flood plains that inevitably place them in harms way. By the 1950s most private companies in America got out of the flood insurance business due to heavy losses from natural flooding. In 1968 the US government unwisely decided to subsidize flood insurance and created the National Flood Insurance Program. One unintended consequence was subsidies encouraged people to continue building in flood plains. As a consequence, by 2014 the program was $24 billion in debt. Recent legislation has attempted to raise flood insurance rates to better reflect the real risks. But raising insurance rates to reflect real risks could force many homeowners into foreclosure and thus any meaningful solution creates a political nightmare. Whatever the political solution, accurate risk assessments require hydrologists and climate scientists to determine the frequency of major flood producing storms over hundreds of years. In Attribution of Extreme Climate Events (henceforth Trenberth 2015) Trenberth suggests extreme storms are more frequent due to global warming. But from a perspective of several centuries, we know flood risks due to hurricanes were greater during the cooler climate of the Little Ice Age (LIA). So how valid and useful is the science of Trenberth 2015?

Textbooks published years before the landfall of Hurricanes Katrina or Sandy prophetically warned, “New Orleans lies below sea level. Should hurricane-driven floods top or break the protecting levees, the city would be inundated with seawater.” And “Large parts of Long Island, New York with its very large population, would be underwater if a major hurricane passed over its western end.” Simply knowing that there was a greater risk of hurricane-induced flooding during the LIA is not really useful for those cities. They already know they are naturally in danger.  Those cities require early warning systems to allow safe evacuation. Accurate early warning requires useful science that can predict the effects of atmospheric circulation and determine storm tracks, storm duration and storm intensity. Oddly Trenberth 2015 argued we should separate analyses of those most useful dynamics and focus on thermodynamics (temperature) because CO2 forced circulation models do a very poor job of simulating those critical dynamic changes. Trenberth 2015 wants to focus on the effect of temperature anomalies in isolation to provide “a better basis for communication of climate change to the public.” But examining temperature anomalies separate from atmospheric circulation changes is dubious science at best and blaming global warming does nothing to improve early storm warnings or accurately assess the frequency of extreme events.

Centennial and Millennial Hurricane Storm Surge

Trenberth 2015 suggested that for Hurricane Sandy, “the subways and tunnels may not have flooded without warming-induced increase in sea level and storm intensity and size, putting a potential price tag of human climate change in this storm in the tens of billions of dollars.” [approaching 50% of the damage].But changes in sea level had little, if any, impact on Sandy’s flooding. His statement may be useful for politicking climate change but does nothing to improve early warning systems. The more useful question to have asked is why was Sandy’s storm surge double that of recent hurricanes, hurricanes that were far more intense but with similar sea levels?

Examining the graphic on storm surge (below) posted by one of Trenberth’s colleagues at the National Center for Atmospheric Research, we clearly see how extreme high water events since 1900 are broken down into contributions from storm surge, high tides and a century of sea level rise. The diagonal orange & white areas represent sea level rise since 1900.  Sea level at Battery Park, NY has risen 11.2 inches over the past 100 years. Half of that rise happened naturally by 1950 - before CO2 had reached significant concentrations - and that natural sea level rise has most likely continued into the present to some degree. Furthermore due to glacial isostatic adjustments, 3 to 4 inches of that relative sea level rise is due to land subsidence on the eastern seaboard. Thus any theoretical contribution from human warming to current sea level is most likely less than 3 inches, and less than 3% of Sandy’s high-water levels. Even if we incorrectly assumed that CO2 caused the entire 1-foot rise in sea level, if we remove that sea level increase Sandy would have still flooded New York’s subways. By blaming global warming, Trenberth 2015 provided nothing useful that would have predicted Sandy’s flooding.

In contrast to Trenberth’s global warming crusade, paleo-climate studies of storm-washed sediments in New York City’s back-barrier marshes show high storm surge was more common when the climate was cooler and sea level was lower. As seen below in Figure 5, coastal flooding similar to Sandy’s happened in 1788, 1821 and 1893. The conclusions from sediment analyses are further supported by historical documentation. The 1893 storm surge was reported to have destroyed Hog Island while driving large boats 100s of feet inland.

Because hurricane caused flooding was more prevalent during the Little Ice Age when Atlantic temperatures averaged 1 to 2 degrees F colder than today researchers concluded, “The frequent occurrence of major hurricanes in the western Long Island record suggests that other climate phenomena, such as atmospheric circulation, may have been favorable for intense hurricane development despite lower sea surface temperatures.” In contrast Trenberth 2015 incorrectly argued analyzing the causes of atmospheric circulation anomalies is not as “fruitful” as analyzing temperatures.

Similarly Liu and Fearn 2000 investigated storm-washed sediments in northern Florida, concluding the region was afflicted with millennial periods of hyperactivity for extreme hurricanes that alternated with a thousand years of quiescent activity. They reported that “no catastrophic hurricane of category 4 or 5 intensity has made landfall in the Western Lake [northern Florida] area during the last 130 year documentary record” but “If future climatic changes, whether or not related to the anticipated greenhouse warming, lead to a return of a “hyperactive” hurricane regime characteristic of the first millennium A.D., then the northeastern Gulf Coast is expected to experience a dramatic increase in the frequency of strikes by catastrophic hurricanes.” Globally other paleo-climate studies found the period of greatest hurricane activity for Australia and the eastern USA both occurred during Little Ice Age times between 1400 and 1800 AD. And in Southeast Asia researchers determined “the two periods of most frequent typhoon strikes in Guangdong (AD 1660–1680, 1850–1880) coincided with two of the coldest and driest periods in northern and central China during the Little Ice Age.”  

Trenberth 2015 wants to re-direct research questions and ask, “Given an extreme storm, how was it influenced by anomalous SSTs?” or “ Was the storm surge worse because of high sea levels?” Based on long-term studies the answer is extreme storms and high storm surge happened more frequently with cooler sea surface temperatures and long before rising CO2. Given that NYC experienced 3 extreme high water levels associated with hurricanes between 1788 and 1893, but only one (Sandy) since then, we can reasonably argue that climate change, whether human-induced or natural, has reduced the threat of high storm surge.

Storm Tracks and Storm Surge

Early warnings and evacuation plans critically hinge on projected storm surge, which primarily depend on the projected storm track.  Perusing hurricane storm tracks since 1850 (illustration below) reveals it was Sandy’s unusual perpendicular approach to the coast that enhanced storm surge. Consider the more intense Hurricane of 1938, which made landfall on Long Island slightly north of New York City as a more intense category 3 hurricane, implying sustained wind speeds between 111 and 130 miles per hour. In contrast Hurricane Sandy made landfall in New Jersey slightly south of New York City as an extra-tropical storm implying winds speeds less than 74 mph. Yet the more powerful hurricane of 1938 only generated maximum water levels at Battery Park, NY of 8.8 feet, and does not make New York City’s top ten high water levels over the past 100 years. It was the difference in storm tracks that determined Sandy’s higher storm surge and higher costs.

Hurricanes produce the highest winds to the right of the hurricane’s direction of travel. Storms travelling parallel to the coastline don’t aim the strongest winds at the coast. When Sandy took a 90-degree turn and travelled perpendicular to the coast, she aimed her most powerful winds at New York City for a more extended period of time as she approached. Due to Sandy’s more eastward position when she started her approach, the fetch was also greater and generated much bigger swells.

The degree of storm surge also depends upon how quickly a storm moves up the coast. Sandy was a hybrid storm that had merged with a cold-core extra-tropical storm typical of winter Nor’easters. Unlike hurricanes that are powered by latent heat from warm sea surfaces, extra-tropical winter storms along the eastern seaboard are primarily powered by the pressure gradient produced by the contrast between the cold continent and warm Gulf Stream. While Trenberth only draws your attention to anomalously warm sea surface temperatures, the east coast was experiencing record cold temperatures that increased the pressure gradient. Forecasters were issuing both blizzard and hurricane watches. Furthermore extra-tropical storms are 3 to 4 times wider than hurricanes, and merging with Sandy produced the hybrid hurricane’s immense size. Extra-tropical storms and their hybrids move much more slowly up the coast than a hurricane, thus the duration of Sandy’s winds generated a much greater storm surge. As seen in Figure 5 above, extra-tropical winter storms (light gray bars) have produced the greatest abundance extreme storm surge. Apportioning partial causation of Sandy’s destruction on global warming and ignoring all else only obscures the critical dynamics required to make early warning predictions based on storm intensity.

Atmospheric Blocking

Due to the frequency of failed forecasts, the public often dismisses media hype about the dangers of an approaching storm, preferring to stay and take their chances rather than needlessly evacuate. Carelessly blaming global warming only adds to the dubious hype and mistrust of useful science. Fortunately the European ECMWF weather models accurately forecasted Sandy’s storm track 8 to 9 days in advance due to a better understanding of atmospheric blocking (in this case the high pressure south of Greenland) and the effects of the jet stream. In contrast, the American National Weather Service’s GFS models initially forecast Sandy to harmlessly head out to the mid Atlantic. Due to such poor forecasting skills, Congress appropriated funds so the NWS could adopt a more accurate weather model. Why did models differ so greatly in forecasting Sandy’s storm track? All the models had access to the same sea surface temperature data, so Trenberth’s temperature anomalies were never a critical factor that could explain model differences.

In fact Trenberth 2015 cited Magnusson 2014 (a paper Trenberth helped craft) in which a ECMWF modeling experiment compared the most recent 20-year average sea surface temperatures with a swath of the Atlantic’s anomalously high temperatures during Sandy’s northward trek. Although that experiment suggested anomalous temperatures could have possibly increased storm intensity slightly, forecasting intensity is still fraught with problems due to the complex contributions from many other variables. More importantly the ECMWF experiment found changes in sea surface temperature had little effect on Sandy’s storm track. Model runs with failed forecasts underestimated the strength of the subtropical high-pressure systems east of the storm track that had kept Sandy from harmlessly veering into the Atlantic. In contrast to Trenberth’s 2015 lament that atmospheric circulation patterns are not robustly simulated by CO2-driven climate models, predicting storm tracks and blocking are the most critical factors for providing early warnings.

Figure 1   Trend in Blocking Days from  Hakkinen 2011

If researchers are interested in a link between Sandy’s storm track and climate change, then a better question to ask would be  ‘have Greenland blocking events been affected by rising CO2 and climate change?” As illustrated in Figure 1 above from the 2011 paper Atmospheric Blocking and Atlantic Multidecadal Ocean Variability, the answer would be there has been no trend in Greenland blocking days (estimates in black and dark blue). Thus a CO2 global warming effect is again unlikely. In contrast, Greenland blocking and hurricane activity are both significantly associated with natural oscillations like the Atlantic Mulitdecadal Oscillation (AMO). The dashed red line represents the AMO and the solid red line represents the detrended AMO.

Klotzbach 2015 has shown that hurricane activity in the Atlantic is highly correlated with the AMO and seemingly independent of climate change. Three peaks of the AMO coincide with 3 peaks of hurricane activity centered on the 1880s, 1950s, and 2005. The oscillation of Atlantic hurricane activity is also illustrated in the Accumulated Cyclone Energy (ACE) index shown below. The AMO appears to be transitioning towards its cool phase now coinciding with a period of below average hurricane activity in the Atlantic since 2013.

Storm Intensity

According to Trenberth’s 2007 article Warmer Ocean’s, Stronger Hurricanes, a one-degree increase in sea surface temperature can increase the winds of a hurricane by one category and he argued global warming will produce more intense category 4 and 5 hurricanes. But Sandy only briefly reached category 3 status as she approached Cuba. Sandy quickly lost intensity after passing over Cuba, devolving from a category 3 hurricane to a mere extra-tropical storm before strengthening again to a weak category 1 hurricane. Clearly the ocean was not warm enough to produce a higher intensity storm that Trenberth and global warming predicted. Or perhaps the dynamic factors that Trenberth downplays had a more powerful part in limiting Sandy’s intensity.

Days before making landfall, due to Sandy’s more westerly storm track, Sandy interacted with an atmospheric trough and its cold Arctic air mass that had dipped down over the eastern USA. The warm-core hurricane named Sandy, eventually merged with a cold-core extra-tropical storm generated by the jet stream. While tropical hurricane intensity is primarily driven by latent heat from warm sea surface temperatures, an extra-tropical storm is primarily driven by baroclinic processes (differences in the pressure gradient) such as the gradient due to the contrast between the warm Gulf Stream and cold continental air mass. As Magnussen 2014 noted, when tropical cyclones and mid-latitude troughs interact to form a “hybrid storm”, it has been found that cyclones are more likely to intensify than weaken. Thus it can be reasonably argued that it was abnormally cold continental temperatures that intensified Sandy.

During a hurricane’s typical cold-induced extra-tropical transition, a hurricane’s size greatly increases as observed in the extremely large radius of Sandy. As reported by Galarneaux 2013, during the transition winds increased by 20% and Sandy’s central pressure dropped to its lowest point of 940 hPa despite travelling over cooler waters. During her “second trough interaction on 29 October, Sandy turned northwestward and intensified as cold continental air encircled the warm core vortex.”  

Everyone agrees that hurricanes require warm waters to form and indeed warmer temperatures can intensify a hurricane. As seen in Figure 10 below, Sandy’s storm track crossed the Gulf Stream (the reddish bands) before making landfall. As she crossed the Gulf Stream, she briefly intensified to a Category 2  hurricane before devolving again to an extra-tropical storm when she crossed cooler coastal waters (in blue and purple). But here again Trenberth’s attempt to separate the dynamics of atmospheric circulation from a thermodynamic impact of higher temperatures would be misleading. It was blocking that forced Sandy to cross over the naturally warm waters of the Gulf Stream. Sandy’s brief increase in intensity was ultimately the result of atmospheric circulation not global warming. Otherwise she would have passed harmlessly out to sea.

The attempts by Trenberth 2015 to suggest global warming has worsened disasters like Hurricane Sandy or the Colorado flooding (discussed in part 2) simply fails to provide any useful science. Trenberth 2015 did not accurately assesses risks or improve early warning systems. It simply reduced climate science to “ambulance chasing” in order to scare up support for his climate change politicking. Previously Trenberth has argued that extreme events such as recent droughts and heat waves worsened due to CO2 warming and despite the fact that climate experts found those events to be within the bounds of natural variability (discussed here). To communicate his brand of climate change, Trenberth attacked those scientists on blogs as irresponsible. And here again Trenberth has hyped global warming links to hurricane destruction in contrast to the opinions of many hurricane experts.

In keeping with the long-term framework required by climate science, hurricane experts like Chris Landsea, the late Bill Gray and Jim O'Brien have consistently reported there are no links between global warming and hurricanes. All the evidence such as the recent lull in Atlantic hurricane activity supports their claims. But despite not being a hurricane expert himself, Trenberth has been grandstanding for a decade to push a climate of fear. After the devastation of Hurricane Katrina, Trenberth convened a press conference to leverage human suffering and blame global warming. Trenberth’s ill-informed bias resulted in hurricane expert Dr. Landsea’s resignation from the IPCC. As Trenberth acknowledged he purposefully convened the press conference to counter publicized reports by hurricane experts that there was no link to global warming. Trenberth defended his conference as necessary to “correct many very misleading and erroneous reports that global warming had nothing to do with the hurricanes in recent times.”  Yet evidence of Trenberth’s links to global warming still remain elusive.

In contrast Landsea’s IPCC resignation stated, “It is beyond me why my colleagues would utilize the media to push an unsupported agenda that recent hurricane activity has been due to global warming. Given Dr. Trenberth’s role as the IPCC’s Lead Author responsible for preparing the text on hurricanes, his public statements so far outside of current scientific understanding led me to concern that it would be very difficult for the IPCC process to proceed objectively with regards to the assessment on hurricane activity.”  Yet here again Trenberth 2015 continues to mislead the public suggesting a storm like Sandy put “a potential price tag of human climate change in this storm in the tens of billions of dollars.”