We all know from basic human experience that the thermometer alone is hardly a complete description of how outside conditions feel at any given moment. The summer sun may be beating down and the air thick with moisture, or a stiff breeze may make a winter's day much harsher than the temperature would suggest. These differences in comfort level can be applied equally to climate as to weather -- for instance, the moist tropical parts of the world are generally perceived as 'hotter' than the dry subtropics even though the latter have the higher temperatures. The combined effect of heat and humidity can be captured in a term called equivalent temperature. Yet there is no universal variable that also takes into account wind speed, solar radiation, and so on. Partly this is for reasons of convenience: data for solar radiation and wind gusts, among others, are rarely saved at weather stations. Various metrics have been derived, some of them proprietary, each with their own advantages and limitations: the AccuWeather 'RealFeel' temperature (with a large number of variables taken from weather models, plus a fuzzy 'perception' factor), the National Weather Service heat index and wind chill, Environment Canada wind chill, the wet-bulb globe temperature, and so on. These are loosely based on physiological studies (e.g. of the amount of caloric expenditure required to maintain homeostasis under a particular set of conditions), with some degree of fudging based on the developer's personal opinions of what constitutes uncomfortable vs. comfortable. In climatology, the wet-bulb temperature is frequently used on the hot end because it can be closely approximated knowing just temperature plus relative humidity or dewpoint.
Intercomparisons between indices have been limited, particularly considering their importance in representing (or at least attempting to represent) the perception of a region's climate to its inhabitants. One effort, now 30 years old, is the "tourism climatic index" (see figure below), taking into account daily-average temperature, humidity, sunshine, precipitation, and wind. While a physiological justification is provided for the inclusion of each variable, the existence of a section called "the issue of the arbitrariness of weights" is reason enough to take any one index with a grain of salt.
My aim was to develop a comfort score that 1. encapsulated key variables (temperature, humidity, and wind), 2. could be calculated from reanalysis datasets to make a gridded product, 3. was simple enough to be computed often, so month-by-month updates could be provided, and 4. reflected common perceptions (in my judgment) of the relative difference in comfort from place to place and time to time. The source dataset is a six-hourly reanalysis, meaning observations are fed into a model that smoothes and interpolates them onto a nice grid. For hot temperatures I used the NWS heat-index formula, and for cold temperatures the NWS wind-chill formula. I defined the optimal adjusted-temperature window as 70 F to 80 F, assigning 'discomfort' weights outside this window based on a modified version of the Anderson and Bell 2009 curve shown in their Figure 2. I compensated for the shifted optimal temperatures by increasing the weight of hot temperatures, and therefore effectively the steepness of the curve above 80 F. My final formula makes equivalent, in comfort terms, a heat index of 90 F and a wind chill of 40 F; a heat index of 100 F and a wind chill of 10 F; and so on.
Once the comfort scores were defined, I calculated climatologies for each month and season across the contiguous US, as well as for each city with a population >100,000 (by triangulating from the nearest three grid points). Annual and seasonal climatology plots are shown in the clickable slideshow below (note the differing scales between plots). The actual values are not important, only how large they are relative to one another. Blues correspond to greater comfort (warmer conditions when it's typically cold, or cooler conditions when it's typically hot), with reds indicating more-intense or -numerous extremes. Temperature is the primary driver, though for example a windier cold snap or a more-humid heat wave could cause a 'red shift' at a location as well.
Another question to investigate concerned the most-comfortable cities (which can be guessed at by examining the above plots). I was curious how these changed from season to season, and if there were significant differences from year to year. The short answer is that, as circuses and state fairs already know (comparing the optimum dates for each city and the dates of the state fairs in the top two figures below show fairly good agreement, indicating that these comfort scores represent roughly what event planners have already taken into consideration), in winter the Gulf Coast is most comfortable, moving up to the Great Lakes & Northeast by June and remaining there through September, then rapidly moving back to the Deep South (4th figure). The most uncomfortable cities (5th figure) in the spring, summer, and fall are in the humid Deep South, moving up to the windy Great Plains/northern Rockies in early winter and over to the Great Lakes in late winter. There is slight variation around these averages on a decadal basis (middle figure), with a warming trend apparent in spring and autumn but not so much in summer or winter. Partly of course this is because there are a limited number of cities used, they are unevenly spaced, weather dominates over climate on short timescales, etc.
But even with these caveats, the numbers match up well with prior expectations: the northward movement of the most-comfortable cities since 1980 has been about 1.5 deg latitude in spring, 0.3 in summer, 1.1 in autumn, and 0 in winter, for an average of 0.73 deg lat. U.S. average temperature has increased about 1.5 deg F during that period; as annual-average temperature in the US drops about 32 F in 20 deg lat, or 1.6 F/lat, all other things being equal the most-comfortable locations should have shifted northward 0.95 deg lat. Not too far off!! In fact, in a post last December about the related "habitability shift", I noted studies that found flora and fauna globally were moving poleward something like 1.7 km/year, which amounts to 60 km or 0.55 deg lat over 35 years. All together, at face value, these numbers suggest raw temperatures are rising fastest, with (for various unexplored reasons) locations of maximal comfort moving somewhat slower, and flora and fauna logically moving slowest of all. ![]()
In this figure and the state-fair one above, the colors are as follows: dark red: late Nov-late Mar --- light red: early Apr/early Nov --- orange: late Apr/late Oct --- light green: early May/early Oct --- green: late May/late Sep --- dark green: early Jun/early Sep --- light blue: late Jun/late Aug --- dark blue: early Jul/early Aug --- dark blue: late Jul
0 Comments
Climate: whatever tends to happen outside, against which we humans can barricade ourselves by building structures that are safe, controlled environments. Right? Not completely... (Sorry for dashing the dreams of the would-be Mars colonists here.) While this definition certainly encompasses most of what is typically meant by the word 'climate', indoor spaces don't exist in a vacuum -- they're more controlled and cordoned off than the free-flowing outdoors, but they can still vary substantially, between regions as well as between individual buildings. These variations in indoor environments are amplified by the large amounts of time that we spend there -- about 87% of our lives, plus another 6% in vehicles. The fact that what we term 'the indoors' is essentially an archipelago of millions of isolated structures, each with its own distinct combination of materials, usage, occupants, contents, and ventilation, in some ways makes indoor climate study as challenging as its outdoor counterpart. First of all, indoor climates demand a different set of tools. While outdoor climate is influenced by familiar factors such as solar radiation, winds, clouds, vegetation, and greenhouse gases, the workings that give rise to indoor climates (air ventilation and materials composition, for instance) are often deliberately hidden from view. Aspects of them are controllable to varying degrees: temperature in a house may be mostly steady during winter, for example, but plunge when power is lost. Just like its analogue in the 'outside' world, this short-lived extreme event can have outsize consequences, from damage to infrastructure to health effects on vulnerable people. Indoor environments are also strongly shaped by the interstitial spaces between outdoor and indoor, whether large enough to accommodate people (i.e. doors) or too tight for the smallest insect. To a large degree, of course, indoor and outdoor climates correlate across space and time. But since the indoors is where most people spent the great majority of their time, whether and how the outdoors comes in deserves considerable interest. This can be modulated by physical aspects of a building, the behavior of its inhabitants, and its immediate outdoor microclimate as noted in a paper from several years ago. For instance: is there A/C and/or reliable heating? Is the building shaded? Are there stoves or filters? How good is the ventilation? Do any of the inhabitants smoke? An extensive indoor/outdoor air pollution study in rural China, contributed to by my friend Ting Zhang, found generally similar or somewhat more elevated levels of pollutants indoors vis-à-vis outdoors. The primary difference was in certain secondary pollutants, which were up to 50% more numerous indoors due to reactions of normal trace air constituents (like ozone) with household chemicals. With its finding of a substantial influence of coal and wood combustion on indoor pollution levels, this study was representative of large parts of the developing world. Given the naturally constricted spaces of buildings, and frequent poor ventilation, smoke from indoor sources is associated with many respiratory and cardiovascular illnesses -- and for tobacco smoke, even "semi-open" spaces trap air enough to be detrimental to health. Particulates emitted from building materials and other indoor sources can also be problematic by accumulating in ways they don't outdoors (even if the emission occurs naturally), as indoors the ratio of pollutant to total air volume can be much, much greater. The obverse is that outdoor sources of pollution have to make it past the barrier of walls and doors, and to fight filtration or other systems. But when those hurdles are minimal, indoor pollution can reach worrisome levels. In an attempt to quantify indoor heat exposure, a recent paper noted that during heat waves indoor dewpoints increase about 0.66 C for every 1.0 C increase in outdoor dewpoint, while for temperature the corresponding figure is only 0.20 C. In a survey of buildings in New York City a wide range of indoor heat indices were found, with a substantial percentage exceeding various heat-advisory levels (see figure above). While the numbers aren't anything outrageous, not taking sufficient precautions or otherwise underestimating these risks because they occur in people's homes or offices rather than out in public -- where environmental factors are more natural to consider -- could result in problems for sensitive groups. In particular, the common advice during periods of hazardous conditions (whether a snowstorm, a hurricane, extreme heat, extreme cold, etc) to 'stay home and stay safe' may need to be qualified to ensure that safety is actually achieved.
Indeed, I would argue that there's no such thing as complete liberation from sensitivity to climate -- it's just a matter of which kind of climate a person chooses to be exposed to at a given time. And we have to keep in mind that our choices have consequences that may not be evident at first blush, on ourselves or on the outdoor climate. But on a more-positive note, indoor climates are arguably more easily controllable (physically and regulatorily), and in the past half-century or so this has given rise to substantial decreases in heat-related mortality, as well as immeasurably improved levels of comfort for those able to afford all of the 21st century's most-advanced luxuries. Whether these improvements in indoor climate (and their attendant economic requirements) can extend to all people, while maintaining some semblance of a balanced and near-natural outdoor climate, is in my mind one of the century's big questions. May: the season of flowers and baby animals and sunshine and snow-season recaps. Over on the Recent Weather page a summary of this winter's total snowfall and snowfall relative to climatology are posted for US cities, as well as the highest snowfall recordings around the world. Available data is spatially inconsistent but does cover at least part of all the major Northern Hemisphere snowy mountain ranges other than the Himalayas (i.e. Japan, Europe, and North America). This year's champion was Alyeska Resort south of Anchorage, which recorded 824" (2093 cm), somewhat above their average of 650". In second place was the weather station on the side of Mt. Rainier which registered 688"; the remainder of the top 10 consisted mainly of the western US, along with Kiroro resort in Japan and Zugspitze in Germany. A warm winter in Europe, especially in the first half, precluded more European ski resorts from joining the list — consistent with warm and wet conditions generally expected there during El Niños. Among US cities, Boulder won by a large margin, 109.7" to second-place Syracuse's 80.3". This was the fourth consecutive year that Syracuse was runner-up, continuing a record five-year dry streak for New York State's snowbelt cities. Given the projections made back in October, this underperformance of the eastern Great Lakes was to be expected, as were several other notable features of this winter's snowfall discussed below. Because of the strong El Niño, average to above-average snowfall was anticipated only in California, Nevada, and to a lesser extent the Southwest into central Colorado, with below-average snowfall arcing from Nebraska across the Great Lakes into New York State. The official NWS forecast along the Northeast coast was torn between the effects of higher precipitation and higher temperatures. This teleconnection-based forecast held true for the most part, with warm temperatures and dry conditions across the northern tier of the country, except for the Pacific Northwest where more precipitation fell than expected. But in Seattle it was never cold enough for snow, causing only the third snowless winter there since records began. Other stations with snowfall considerably below average included Billings (located in the dry spot on the October NWS forecast), Anchorage, and a small pocket of eastern Kansas and western & central Missouri. Anchorage is an interesting case as it is just 40 mi from the aforementioned Alyeska Resort, and only 2500' lower. However in a warm winter, and with the resort facing the ocean with the city in its rain shadow, these little twists of geography make all the difference. After a quick start (snow on the ground continuously from Oct 30 to Dec 28), Anchorage recorded just one day >=1 deg F below average from Jan 1 through Apr 30. Unlike Alyeska, Anchorage snowfall shows a clear decreasing trend with higher temperatures (see below), and that played out again in this El Niño-dominated year. Kansas and Missouri experienced a wet December but a dry winter after that, ending up as one of the least-snowy winters on record for Wichita, Kansas City, and Columbia, MO. They were expecting a somewhat snowier-than-average winter based on the El Niño teleconnection-based forecast, but the average storm track was further north than usual for an El Niño event, with (directly to their north) Sioux Falls, SD receiving its 5th-most snowfall. The Front Range in Colorado benefited from several spring storms, while the Northeast from Washington to Boston was subject to one big storm on Jan 22-24, and a smaller one in March in New England. Even for a region in which snowfall is concentrated in a handful of events each year, the dominance of that single event is remarkable: in 24 hours, 27.3" fell in Central Park, the largest single-day total on record there and a full 83% of the seasonal total of 32.8". This concentration far outstrips any previous storm/season ratio, as shown in the figure below. All this thinking about snow made me wonder what the observed snowfall trends are like across the US. After all, in many regions there is something of a tug of war going on, often between temperature and precipitation; between greenhouse-gas warming and aerosol cooling; and/or between long-term warming and natural interdecadal variability. The ultimate effect in a particular place typically depends on its idiosyncratic climate characteristics. For example, in the Great Lakes warming has increased water temperatures and decreased ice cover, increasing the potential for precipitation, but the increase in temperature has not yet been enough to convert much of the snow into rain -- resulting in greater lake-effect snowfall. (This is akin to observed increases in snow cover in the high elevations of the Hindu Kush, where precipitation is increasing but temperatures are still well below freezing.) At some indeterminate point in the future, scientists expect the balance to tip and snowfall to begin decreasing.
The trend in total seasonal snowfall over the period of record for each city (ranging from 50 to 104 years) is shown in the first figure below. At the 95% confidence level, increases are found throughout the Great Lakes as well as in a few other cities such as Fargo, while only one decrease is found, in Denver. Examining the plots for individual cities it seemed as if the increase might be in large part explicable by an increase in interannual variability in recent decades, and the second figure below partially confirms this. I defined variability as the standard deviation of a 20-year period of annual total snowfall, moved this window forward one year at a time, and finally computed a trend over these standard deviations. Looking now at the 90% confidence level, there are indeed increases for some of the same Great Lakes stations, as well as some stations on the Northeast coast, Fargo, and Amarillo. Milwaukee and Lexington, neither of which is substantially affected by lake-effect snowfall, exhibit decreases in variability. Speaking broadly, then, while there is some hint of increases in variability in parts of the Great Lakes and along the Northeast coast, the observed (mostly positive) changes in snowfall over the 20th and early 21st centuries cannot largely be explained by it. The climate has always been variable — it's more that now (to reference last month's post) we as a society have a greater expectation of insulation from nature's vagaries, and the ability through instant communication to pay more notice to extremes no matter where they occur. This past winter is an exemplar of that — while notable in the ways described above, none of the 73 cities tallied experienced a record high or low in snowfall, while because there are 66 years analyzed the most likely situation would have been one record high and one record low. Having strong computational models and knowledge of correlational relationships like El Niño teleconnections allow us to make better predictions, as borne out in October's largely verified forecast, but that's not the same as domesticating it — the climate is still wild at heart. Note (6/15/16): This post proved prescient, as an article has just come out making more explicit the link between greater global economic connectivity in the last several decades and greater systemic vulnerability to climate extremes (in this case, heat stress). With the amount of money at stake and the size of the uncertainties involved, not to mention the natural counterbalancing that will occur as the benefits of free trade are defended, surely more such articles picking apart different aspects of this economic-climatologic intertwining are on the way. One of the many paradoxes of modern society is the seeming disconnect between the exponential increase in information and technological prowess on the one hand, and the decidedly slower increase in living standards, freedom, and social stability on the other. Move down a peg and there is another disconnect, between that slow methodical improvement of the world and people's pessimism about its direction — in almost every country the majority of poll respondents report they perceive the world as getting worse, in many cases by enormous margins. Both of these points were brought up very thoughtfully in a climate context in a January blog post by William Hooke, and I thought them so valuable that they were worth expanding upon here. As Hooke writes, modern society is built on margins — in finance, commerce, and even in love we are living with the knowledge that our competitors are very similar to us and it's that little bit extra that's enough to push us over the edge. Trade (of stocks and goods), just-in-time manufacturing, and online dating are just a few of the ways that efficiency is wrung, and advances are made, out of reducing the margin/processing cost to as small a value as possible. These advances are enabled by technology and global interconnectedness, but at the same time this system poses dangers for the exact reason that everything is linked together. Like dominoes that were once far apart but are now pushed close together, the elements of this system are worryingly interdependent, where the improvements in domino sturdiness (i.e. greater agricultural productivity, technological redundancy, gains in predictability of disasters) are offset by the fact that the consequences if they do fall are much more catastrophic and widespread. In other words, I argue that in many ways most people are no longer truly prepared for uncertainty because (despite our collective pessimism) we tend to conceive of large-scale disasters as a thing of the past. For instance, governments are no longer capable of, much less inclined to, stockpile food for the proverbial '7 years of famine', something only conspiracy cranks and fatalists are worried about. And this is not an issue unique to climate change — in a stationary climate they could also occur, though their probability would be somewhat lower considering that there have been observed increases in most kinds of extremes around the globe in recent years. Global production of commodities like food has risen exponentially over the past several generations, and in proportion we have lost a visceral fear of threats like imminent starvation; rather, people's pessimism seems to be based more on the grounds of a perceived risk of decreases in morality and leisure. Many countries have so-called rainy-day funds, but ironically in terms of physical necessities like water the trade-based global economic system has worked well enough (for those who can afford to purchase commodities at the going rate, anyway) for so long that an assumption seems to have set in that all the problems caused by 'small' idiosyncratic national-level sources of uncertainty — political, social, climatological — can be addressed with technology or trade, not giving much credence to potential systemic ones. These latter might take the form of simultaneous extreme heat or drought cutting yields in multiple of the world's breadbaskets at once, or disease racing through the genetically near-identical population of Cavendish banana trees, or a weak monsoon wreaking havoc on rice-growers in both India and China. We don't know exactly how likely such situations might be, only that they would be devastating and therefore should at least be thought about, if not actively insulated against. In some aspects of modern life, perhaps primarily those that are generally conceived of as risky, risks are calculated, contingencies are prepared for, and arguments are made as to whether preparations are sufficient. In others, like the environment, certain risks are also estimated as best we can with models, theory, etc. — but at the same time there is the "Titanic problem": that a wide array of climate risks can be well-estimated and prepared for in and of themselves, but, like the sealable compartments that the Titanic contained, the possibility that they may occur in devastating simultaneity is rarely accounted for. These caveats, of course, are properly read as a subtext to the recognition that global trade and margin-based economics have been essential to the productivity gains of the Industrial Revolution and thereby to the modern way of life. Indeed, it is almost because the system has worked so well (again, excepting those who were shut out of it for lack of funds or political access) that a certain level of complacency has developed. In a few short generations the popular mood can swing from existential fear of something to complacency, which can further veer over to rebellion, such as with the current anti-vaccination movement, spearheaded by people who surely do not recall the times when 21,000 Americans were paralyzed per year by polio. Shaking off this sense of complacency and narrow conception of climate- and weather-related risks, and adopting a global-scale perspective of them as befitting a global-scale society, is essential to properly and efficiently mobilizing resources to prepare for them. Marginal thinking has enabled the achievement of dazzling heights of productivity and inventiveness, and so in some ways has given us the tools to hasten its own demise — having the resources to liberate ourselves from the paradigm of thinking that it's the solution to all the world's problems.
|
Archives
September 2023
Categories |