Atlantic multidecadal oscillation

• 1. Introduction

  ›

  The Atlantic Multidecadal Oscillation, known as the AMO, describes a slow rhythmic warming and cooling of the North Atlantic Ocean surface that plays out over decades. It touches droughts in the American Midwest, hurricanes in the tropics, rainfall over the African Sahel, and even snowpack in the Alps. The term itself was coined in a 2000 telephone interview when climate scientist Michael Mann spoke with journalist Richard Kerr, as Mann later recounted in his 2012 book The Hockey Stick and the Climate Wars. That casual conversation gave a name to a pattern researchers had been chasing since the 1980s. Yet decades later, Mann himself would publish research questioning whether the AMO exists at all. How did one of climate science's most influential concepts become one of its most contested? And what does it mean for every forecast tied to it, from hurricane season outlooks to long-range drought planning?

• 2. A Pattern Takes Shape

  ›

  Folland and colleagues first surfaced evidence of a multidecadal climate oscillation in the North Atlantic in work published during the 1980s. The signal they identified was striking: the North Atlantic appeared to warm and cool in long, slow pulses. Schlesinger and Ramankutty made that oscillation the sole focus of a 1994 study, pulling the phenomenon into sharper scientific focus.

  Based on roughly 150 years of instrument records, researchers identified a quasi-periodicity of about 70 years. Warmer phases appeared from around 1930 to 1965 and again after 1995. Cooler stretches ran from roughly 1900 to 1930 and from 1965 to 1995. Enfield and colleagues projected that the warm regime beginning around 1995 would persist at least until 2015 and possibly as late as 2035, with a peak projected around 2020. But building a reliable index from those observations proved harder than naming the phenomenon.

• 3. The Index Problem

  ›

  Trenberth and Shea proposed subtracting the global mean sea surface temperature, averaged between 60 degrees North and 60 degrees South, from the North Atlantic average to isolate the AMO signal. Ting and colleagues pushed back, arguing that the forced warming pattern is not globally uniform. They separated forced and internally generated variability using a signal-to-noise maximizing technique called EOF analysis.

  Van Oldenborgh and colleagues took a different approach: they averaged sea surface temperatures over the extra-tropical North Atlantic to reduce El Nino influence, then subtracted a regression on global mean temperature. Guan and Nigam removed both the non-stationary global trend and Pacific natural variability before applying their own EOF analysis to the remaining North Atlantic signal.

  The competing methods do not just disagree on arithmetic. The linearly detrended index suggests that warming at the end of the twentieth century was split roughly equally between external forcing and natural variability. Other methods conclude that a large portion of that warming was externally forced. In 2017, Frajka-Williams and colleagues pointed out that recent cooling in the subpolar gyre, warmth in the subtropics, and cool anomalies in the tropics had created a north-south gradient that no single AMO index value could capture.

• 4. Droughts, Storms, Sahel

  ›

  Two of the worst droughts of the twentieth century fell squarely within a positive AMO phase lasting from 1925 to 1965: the Dust Bowl of the 1930s and the widespread drought of the 1950s. During warm AMO phases, droughts in the US Midwest and Southwest tend to be more frequent or prolonged. Florida and the Pacific Northwest run counter to that pattern, receiving more rainfall when the AMO is warm.

  Rainfall over northeastern Brazil and the African Sahel also shifts with AMO phase. Paleoclimatic studies confirmed this Sahel pattern across the past 3,000 years, showing increased rainfall during warm AMO phases and decreased rainfall during cool ones. Climate models suggest that a warm AMO phase strengthens summer rainfall over India as well. Spring snowfall over the Alps and glacier mass variability also track with AMO shifts through changes in atmospheric circulation. At least twice as many tropical storms can mature into severe hurricanes during warm AMO phases compared with cool phases.

• 5. The Hurricane Debate

  ›

  A 2008 study matched the Atlantic Multidecadal Mode against hurricane records from the HURDAT dataset covering 1851 to 2007. It found a positive linear trend for minor hurricanes in categories 1 and 2, but that trend disappeared once the authors corrected for undercounted historical storms. The study concluded that any greenhouse gas-driven increase in hurricane activity was currently obscured by the 60-year quasi-periodic cycle.

  Mann and Emanuel reached a different conclusion in 2006, finding that anthropogenic factors drove long-term warming in the tropical Atlantic and that there was no apparent role of the AMO. Then in 2014, Mann, Steinman and Miller argued that past methods for estimating the AMO had inflated its amplitude and skewed its phase. They wrote that the true AMO signal appeared likely to have been in a cooling phase in recent decades, quietly offsetting some anthropogenic warming rather than adding to it.

• 6. Prediction and Its Limits

  ›

  Instrument records spanning roughly 130 to 150 years provide too few data points for conventional statistical methods to pin down AMO periodicity with confidence. Enfield and Cid-Serrano extended the analysis using a 424-year proxy reconstruction. Their histogram of phase-crossing intervals showed the highest frequency of regime shifts occurring every 10 to 20 years. The cumulative probability of a shift within 20 years or less was about 70 percent.

  No deterministic prediction method has yet demonstrated it can say when the next shift will arrive. Computer models capable of forecasting El Nino months in advance remain far from reliable for AMO timing. A 2017 study predicted a continued cooling trend beginning in 2014, noting that the pattern of temperature anomalies had taken on a distinctive tripole shape: cold in the subpolar gyre, warm in the subtropics, and cool over the tropics. That pattern, the authors warned, may increase atmospheric instability and storminess in ways the standard AMO index cannot represent. Water managers and agricultural planners in drought-prone regions depend on long-range AMO projections for infrastructure decisions, which means the stakes of that measurement gap are real.

• 7. Does the AMO Exist?

  ›

  In 2021, Mann and colleagues published what they called "the most definitive evidence yet that the AMO doesn't actually exist." Their study showed that the apparent periodicity of the AMO over the past millennium was driven by volcanic eruptions and other external forcings, not by an internally generated ocean cycle. Models showed no oscillatory behavior on time scales longer than El Nino Southern Oscillation. The AMV signal, they concluded, was statistically indistinguishable from red noise, the baseline test used to check whether a genuine oscillation is present.

  The man who named the AMO in a phone call in 2000 had, two decades later, provided the evidence that the name may have attached itself to something that was never truly there. The controversy matters well beyond terminology: if the warming and cooling attributed to the AMO are actually responses to volcanic forcing or greenhouse gases, then every correlation built on AMO phase needs to be re-examined against what actually caused the sea surface temperatures to change. Mann, Steinman and Miller's 2014 finding that standard AMO estimation procedures yield an inflated amplitude and biased phase remains the sharpest statement of what that re-examination would involve.