by Judith Curry
Nature’s exuberant smashing of daily high temperature records [in the U.S.] in recent weeks can only be described as “Meteorological March Madness”. Conditions more fitting of June than March prevailed east of the Rocky Mountains since the start of the month. NOAA’s National Climate Data Center reported that over 7000 daily record high temperatures were broken over the U.S. from 1 March thru 27 March.
My favorite extreme weather attribution group at the NOAA Earth System Research Laboratory (led by Martin Hoerling) has a new report on the weird weather this past March in the U.S. To see all the cool graphics, you will have to link to the report. Here are some excerpts:
1. What were the meteorological conditions associated with the heatwave?
It is first useful to place the heatwave into a Northern Hemisphere context . The heatwave was clearly regional in scope and was not part of a pervasive hemisphere-wide warm regime. Rather, widespread cold conditions at the same time occurred over the western U.S., western Canada, Alaska, eastern Asia, and southeast Europe. In a similar context, it is useful to recall that the prior month of February 2012 which was generally very warm over the U.S., was very cold over Europe, and global land temperatures ranked 37th warmest in over a century, representing the coolest February since 1994 (http://www.ncdc.noaa.gov/sotc/global/2012/2).
Over the Eastern United States, select time series of daily temperatures reveal the timing and magnitude of the maximum warming. A peak warming in the upper Midwest spans roughly a 12-day period from 12 – 23 March.
2. What physical processes contributed to the heatwave’s magnitude?
An appreciable fraction of the magnitude of the warming is thus reconcilable with a simple transport of heat from Gulf Coast region northward.
The 850mb and 500mb anomalies during 12-23 March 2012 are characteristic atmospheric features of variability, ones that are effective in inducing poleward heat transport and strong temperature variability over the northern U.S. They were, however, extreme circulation anomalies in terms of magnitude and perhaps persistence.
Have these features of the time-averaged atmospheric circulation become more “energetic” in the sense that the current 2012 unusual intensity of atmospheric dynamics might be related to an increase in variability over time? Perhaps the explanation is “weather weirding”, “weather on steroids”, or an extreme non-linearity to assert how the atmosphere may be responding to increases in greenhouse gases? Such explanations can be tested with data and modeling experiments. Here we present some preliminary analysis of March daily and monthly means. In summary, our diagnoses provided below show no support for the supposition that there has been enhanced variability, at least to date for the particular variables of immediate relevance for the March 2012 heatwave. Neither the monthly or the daily data reveal appreciable increase in the variability of March weather, though this does not negate the possibility that such patterns may change in the future. And, while finding no evidence for a change in variability, it is nonetheless evident that the anomalies in 850mb temperatures, 850mb winds, and 500 mb heights related to the heatwave were quite likely of historic proportions for this time of year, with departures during 12-23 March 2012 on the order of 5 times the magnitude of typical variability.
As shown by our analysis of observational data, an explanation that this heatwave was an outcome of a strong nonlinear feedback associated with a climate change induced reduction in snow cover or dry soil conditions must be rejected based on evidence and physical understanding regarding conditions associated with this particular heatwave event.
First, as noted above, much of the region which experienced record heat does not normally have snow cover in March, thus this mechanism does not apply for most of the area that experienced record March heat.
Second, the North American trend in March snow cover has been upward, not downward. The principal decline in snow cover extent emerges in late spring, when the climatological snow extent pushes well north into Canada. March and May snow cover changes have been materially different from each other over North America, and indeed of opposite sign in recent decades.
So, while a quantification of the snow impact on the current heatwave magnitude cannot be precisely discerned without further diagnosis, it is very likely that changes in snow were a response to the heatwave, not a cause for its extreme magnitude.
3. Was this extreme March 2012 U.S. heatwave event anticipated?
Here we explore the evidence for a substantial human influence on the March 2012 heatwave, both on its magnitude and on the probability of such an extreme event occurrence. There is no doubt that there exists an influence of human-produced greenhouse gases on evolving weather and climate conditions, as the IPCC reports have clearly enumerated. Yet, while acknowledging that climate change plays a role in every weather event begs the issue of the direction of that impact and the magnitude of its effect. Unfortunately, when claims are made that a event “would not have happened without human produced greenhouse gases”, the incorrect conclusion is all too often drawn by the public and media that an event was caused, in some de novo manner, by human climate change. The pitfall of such framings is apparent from the well-known nature of chaotic weather and climate systems. In that paradigm of a nonlinear system, no event today or tomorrow would have happened owing to the smallest flap of a butterfly’s wings. But is there any predictability, and in particular, could the March 2012 heatwave have been anticipated?
Regarding the connection of greenhouse gases and extreme weather, the matter of predictability has often been cast in the context of changed probabilities. As employed in the study of the 2003 European heatwave (Stott et al. 2004) and also the 2010 Russian heatwave (Dole et al. 2011), the question posed is how increased greenhouse gases changed the percent probability of reaching a particular record (or event) threshold. Of course, this is only one aspect of early warning. In the case of the current heatwave, we wish to understand the extent to which GHG forcing offered any of the specificity that characterized the event, namely that it happened in March, that it happen in 2012, that it happen in the Upper Midwest/Ohio Valley region, and that the event had such extreme magnitude.
What is also important to recognize is that the GHG warming signal for March 2012 offers few of the attributes that are required of a meaningful early warning. First, it fails to anticipate the event happening in 2012, since a similar warming signal existed in 2010 and 2011 (and is again anticipated in 2013), Second, it fails to anticipate the particular monthly occurrence in March, since a similar warming signal exists throughout the annual cycle. Third, it fails to anticipate the specific location of the 2012 heatwave over the Upper Midwest/Ohio Valley region since the magnitude of the GHG signal is not materially different for the western U.S., central U.S., and eastern U.S., or for that matter for Alaska, and Asia which, during this period, experienced severe cold. And, the GHG warming signal fails to explain the extreme magnitude of the heatwave event, which achieved daily departures of +20°C, or about 20-fold greater than the estimated background warming signal. Our physically based analysis explored the possibility for strong nonlinear amplifying feedbacks or highly nonlinear dynamical sensitivity to GHG forcing, and these models which represent such process in their integration of the coupled climate system, are consistent with that empirical assessment that such processes were unlikely of primary significance over the U.S. during March 2012.
The weak overall contribution of GHG warming to the magnitude of the March 2012 heatwave notwithstanding, a signal of about +1°C warming appreciably increased the odds of a record March heatwave occurring. Such a signal corresponds approximately to a 0.5 standardized shift of the probability distribution toward warmer conditions. There are, however, statistical challenges in estimating how such a shift in distributions would alter extreme event odds, especially of the intensity observed in March 2012 whose magnitude was likely on the order of 4 – 6 standard deviations. The limited length of the available data records makes it difficult to discern the nature of the frequency distribution from which our extreme March 2012 heatwave event was drawn. Compounding the difficulties is the extreme sensitivity in estimating changes in probabilities of threshold exceedences to the assumed background distribution. The typical assumptions of a normally distributed (Gaussian) temperature distribution does not hold for daily values in particular. Nor do they generally hold for monthly averaged data when considering the extreme tails of distributions. This issue of estimating reliable statistics of extreme, rare events continues to be a matter of active research. The assumption of Gaussianity is more defensible, however, if considering a threshold of more moderate magnitude. Here we consider the changing odds of exceeding a threshold for a 2 standard deviation departure (~4°C) March temperature condition. This is roughly akin to a 1 in 40 year event magnitude. For a 0.5 standardized anomaly of GHG warming (+1°C), the exceedence probability increases roughly by 50%, and somewhat less than a doubling in the probability.
A black swan most probably was observed in March 2012 (lest we forget 1910). Gifted thereby to a wonderful late winter of unprecedented balmy weather, we also now know that all swans are not white. The event reminds us that there is no reason to believe that the hottest, “meteorological maddest” March observed in a mere century of observations is the hottest possible. But this isn’t to push all the blame upon randomness. Our current estimate of the impact of GHG forcing is that it likely contributed on the order of 5% to 10% of the magnitude of the heat wave during 12-23 March. And the probability of heatwaves is growing as GHG-induced warming continues to progress. But there is always the randomness.
Robin Webb of NOAA sent me this figure of the 1910 heat wave in the midwest U.S. March_1910_20cen (sorry i couldn’t insert it into the post, you can click on it to open).