Where Is the World’s Best Weather? (2020)

Posted by admin on July 30, 2021

I was thinking about becoming an expatriate (“expat”) — about living abroad, at least for a while. The question was, where? I had already done some work to figure out which places were most affordable; now I wanted to know where I could find the best weather.

(Note: in the process of developing this post, I concluded that weather would be only one factor affecting my planning, and not necessarily the most important one. Thus, I did not finish fully exploring weather to the extent originally anticipated. As such, the best use of this analysis may be to identify criteria of interest when considering the weather of specific places. In a few spots, this incomplete writeup may also have minor problems of coherence.)

What Weather Is Best?

In trying to find the world’s best weather, there were some things I had to sort out. One was that you can have it all. You can have four seasons with a cold winter, and you can also have nice, warm, sunny January weather. You just can’t have it in the same place. You have to get in your vehicle and drive, or climb on the plane and fly. To enjoy this flexibility, it helps to live in a geographically and climatically diverse country like the U.S. — or (judging from Google Earth) Argentina, or perhaps Italy; or in a place where the mountains rise up out of the sea to put those cold and warm spots right next to each other, like the west coast of South America.

But I wasn’t planning on being a snowbird — spending winter months in a warmer location, that is, and then going somewhere else in the summer months. Although I wasn’t sure what the research said, I had heard that people who are gone for a substantial chunk of the year may find themselves socializing primarily with other snowbirds, if any are available and interested in socializing; they may not engage very deeply with society and community in the places they live. Snowbirding can also be difficult for those who want the advantages of a resident’s visa in their target countries: apparently there could be legal limits on how long one can be gone, at least in the early years. Of course, there can also be travel expenses in shlepping back and forth between locations, as well as the higher costs of maintaining two properties, keeping full sets of possessions in two places, and/or being limited to short leases.

The mission, here, was to identify a destination where I would want to stay. Not that I wouldn’t consider going somewhere else sometime; I just wanted to make my first choice the best one. Which raised the question: in weather, what is “best”?

I decided there was a good answer to that question — that it wasn’t necessarily just a matter of opinion — and that it mattered. Lucas and Lawless (2013, misconstrued by Darack, 2014; accord Feddersen et al., 2013) found that weather does not make a difference in the average person’s degree of life satisfaction. But that might not apply to people for whom daily exposure to the outdoors is an important element of preferred lifestyle, or whose social interaction is affected by weather conditions (e.g., Perry, 2014) — including particularly snowbirds (Simpson & Siguaw, 2012). Considerable research has demonstrated that exposure to the outdoors tends to have a positive impact upon psychological well-being (e.g., Bratman et al., 2015; Buchecker & Degenhardt, 2015; Whear et al., 2014; see also Dolan & Laffan, 2016). So if good weather got a person outside more frequently, then maybe the pursuit of good weather would help. Along these lines, a search led to several additional interesting articles:

In a study of office workers in China, consistent with other research, Cai (2013) found that productivity and satisfaction with the location of one’s workstation generally increased with proximity to windows (but declined for those sitting at desks immediately next to windows) and were higher on east, south, and west (i.e., at least partly sunny) sides of buildings than on the north side.
Glimcher and Tymula (2015) cited research finding that older people are more vulnerable to weather and climate changes; that people are more risk-averse on cloudy days, when preceding days were not cloudy; that luminance (i.e., intensity and duration of sunshine) promotes optimism but also inconsistency, especially when luminance is higher than on previous days; and that bad mood (i.e., pessimism) improves analytical performance and memory (suggesting improved consistency).
Chang (c. 2016) alluded to (but failed to specifically cite) research suggesting that time spent outdoors improves health in several ways: reduced stress and depression levels; improved creativity and productivity; improved ability to concentrate; Vitamin D absorption; for older people, improved mobility and ability to participate in daily activities, with reduced complaints of aches and of sleep problems; improved social skills for older dementia and stroke patients; strengthened immune systems; and improved eye health, quality of sleep, and self-esteem. ASC (n.d.) echoed those points and added (with citations) that indoor air tends to be significantly more polluted than outdoor air, and that outdoor activity increases mobility, energy level, and happiness for people generally (i.e., not just older people).

In short, these sources suggested that, as one might expect, exposure to the outdoors — particularly outdoors activity — is physically and mentally healthy. Therefore, I decided that the best weather was the weather that would most encourage a person to go outdoors and stay outdoors. For that purpose, key weather elements would be sunshine, warmth, and dryness — calling to mind the Dreary Weather Index offered by Brian Brettschneider:

Immigration from south of the border may have accounted for some of the significant population growth in cities in Texas, Arizona, Florida, and southern California. But it did not account for population declines in Rust Belt cities from Milwaukee to Rochester. For a century at least, Americans had been flocking to retirement destinations in southern and southwestern states. While some internal migrants may take account of tax breaks, there was no comparable flow toward South Dakota or other low-tax states in the north (see e.g., Conway & Rork, 2016, pp. 1012, 1021; Mazerov, 2014). Likewise, from what I had seen in my browsing, the preferred countries for expat retirement tended to be closer to the equator: Mexico, for example, and Thailand — not, for the most part, Canada or Sweden, despite their potential social, cultural, and/or linguistic advantages. American retirees willing to take a chance seemed much more inclined to consider affordable Portugal or Panama than affordable Estonia or Siberia.

Granted, not everyone will enjoy the same climate, or the same weather events. Some of that is due to individual physical characteristics: metabolism and other aspects of a person’s body will influence how s/he reacts to warmer or cooler temperatures. Some is due to the conditions in which one encounters the weather. What is warm to a person doing physical work in the sunshine can be cool to a person sitting in the shade. Some is due to personal psychiatric composition: for example, while Seasonal Affective Disorder is commonly understood as the experience of depression during dark, cold winter months, for some it is summertime that brings on seasonal depression.

Again, it wasn’t that people don’t love Vermont or the Pacific Northwest, and it wasn’t that I would want to spend the rest of my days locked into a desert hellhole. This was just an attempt to identify the kind of weather — climate, actually — that would be most conducive to a tolerable retirement. Consistent with various sources discussing which U.S. cities had the best weather, I saw the paths worn in the freeways by the Americans (and especially the retirees) before me. I had lived in a variety of places within the United States, from Northeast to Southwest. I knew what kind of weather it took for me to spend hours each day outdoors, or at least with the windows open. If I had to choose one place to live in, within the U.S., a place in the Southwest, or at least somewhere in the South, would be the safest bet.

I did think that, given a choice, most people would prefer some seasonal variation. A warm to mild winter might be OK — might even be ideal — as long as it did not interfere with the priority of being able to go outdoors easily and stay out there for long stretches.

Preferred Climate Types

Ever since my elementary school days, I had seen maps showing different climate zones around the world, such as this example from Wikipedia, apparently taken from the U.S. National Oceanic and Atmospheric Administration (NOAA):

Now it turned out that the climate map offerings had improved since my grade-school days. My search led to a number of interesting interactive (e.g., zoomable) maps. Among those I examined, MeteoEarth offered a very intuitive globe, with options to zoom in (roll the mouse wheel), move around (left-click-drag), and view any combination of temperature, precipitation, cloud cover, wind, and other elements, now and for the next 24 hours (so as to view overnight changes). National Geographic also offered a globe with better viewing of the overlap of climate zones and national boundaries, and with an option to choose different base maps (e.g., topographical, satellite photo); at present its other features seemed to be undeveloped or dysfunctional. NOAA offered an interactive map that was not zoomable, but that did allow me to click on a specific climate zone for a paragraph’s worth of generic information about that type of zone (e.g., “mild to warm summers and cold winters”; “Inland locations may also have more precipitation during the summer months”).

It seemed that what people liked about the weather in the American South or West could be summarized in terms of climate types. Wikipedia explained that there were different climate classification systems — and that, of these, the Köppen system was the most commonly used. The World Meterological Organization (WMO) explained several different kinds of averages used in meterology and pointed me toward a choice among several world data centers.

Armed with that background information, I returned to the question of what kind of climate people liked, in the southern and southwestern U.S. Closer examination of interactive and regular maps, displaying Köppen climate zones, instructed me that there could actually be multiple climate zones within the same metropolitan area. From what I saw, it seemed the most popular Köppen climate types would include Cfa (found in, for instance, the U.S. southeast of 38°N 100°W), Csa (e.g., Sacramento, Anaheim), Csb (e.g., San Francisco, Long Beach), BSk (e.g., Albuquerque, Irvine, coastal San Diego), and BWh (e.g., Phoenix).

I wondered whether those were also the preferred climate types abroad. Worldwide, 11 different sources (i.e., USA Today, TMW, Huffington Post, Traveller, Weatherwise, GoTopTens, TripsToDiscover, AntonioGarzon, Earth & World, Tourist Maker, Mother Nature Network) listed cities with supposedly perfect weather. Among the many cities named by such sources, there was agreement by at least three sources on these cities, in descending order: San Diego, CA; Las Palmas (in the Grand Canary Islands); Malaga (or the Costa del Sol), Spain; Sydney (or Adelaide), Australia; Sao Paulo (or Rio de Janeiro), Brazil; Honolulu (or other Hawaii); Nice (or Marseille), France; Loja, Ecuador; Kunming, China; Port Elizabeth (or Durban or Cape Town), South Africa; and Medellin, Colombia. Several coastal cities north of San Diego, between Oceanside and San Luis Obispo, were also mentioned. I chose Long Beach, from the previous paragraph, to represent them, so in that sense Long Beach was also implicated by several of these travel writers.

I was not sure what was the best way to look up the Köppen climate codes for those places. I posted a question on Quora, asking if there was a list or spreadsheet offering that information. That question remained unanswered; eventually I answered it myself. The answer I gave was that I had found a downloadable spreadsheet offered by SG Kinsmann at the 400 Cities Digital Nomad Index, providing the Köppen codes for 373 cities. To check that spreadsheet, I manually looked up the climate types for the places listed in the preceding paragraph. To do that, I thought at first of using a map, but the available maps did not look very specific and would seem to require judgment calls among several climate types, in and around unfamiliar places. The results of my lookups, and the sources I used, were as follows:

San Diego: BSh (Wikipedia), BSk (Weatherbase, ClimaTemps, Climate-Data)
Las Palmas: BWh (Wikipedia, Weatherbase, Climate-Data)
Malaga: Csa (Wikipedia, Weatherbas, Climate-Data)*
Sydney: Cfa (Wikipedia, Weatherbase, Climate-Data)*
Sao Paulo: Cfa (Wikipedia, Weatherbase)*, Cfb (Climate-Data)
Honolulu: BSh (Wikipedia), As (Weatherbase, ClimaTemps, Climate-Data)
Nice: Csa (Wikipedia, ClimaTemps, Climate-Data), Csb (Weatherbase)
Loja: Cfb (Weatherbase, ClimaTemps, Climate-Data)
Kunming: Cwb (Wikipedia, Weatherbase, Climate-Data)
Port Elizabeth: Cfa (Weatherbase), Cfb (Wikipedia, Climate-Data)*, Csb (ClimaTemps)
Medellin: Af (Wikipedia, Weatherbase), Am (Climate-Data)

The climate codes in this list captured all of the codes that I had identified in popular southwestern American cities (above). The list demonstrates that sources disagreed, as in the U.S. cities I had already checked (above). The confusion seemed to be worse on islands. The Canary Islands apparently had a number of different climates, mostly BWh but also including BSh and Csb among others, and Hawai’i magazine (Paiva, 2015) reported that Hawaii had ten different kinds of climate.

At any rate, asterisks (*) indicate which climate code was designated, in the Kinsmann spreadsheet, for the foregoing listed cities. Kinsmann differed from all of the foregoing sources in ways that were at least partly supported by Plantmaps, in giving San Diego a Csa, and giving Honolulu an Aw. Kinsmann also gave Medellin a Cfb — reasonably, according to Wikipedia — and had no data for Las Palmas, Nice, Loja, or Kunming. I concluded that Kinsmann’s spreadsheet was somewhat incomplete, and was not highly consistent with the foregoing sources, but that it did not differ ineptly, and therefore would probably be adequate for general purposes.

There were 29 codes in the Köppen system. The foregoing list of what I will call “perfect weather cities” from around the world, including the examples of desirable U.S. cities, contained 11 of those 29 (i.e., Af, Am, As, BSh, BSk, BWh, Cfa, Cfb, Csa, Csb, Cwb). While not all of those 11 clearly applied to the perfect weather cities, it appeared they were close enough to have potential application. Meanwhile, the list of 29 would have to be shortened for present purposes, as a number of other Köppen codes applied only to extreme weather areas that would not be relevant here. In other words, the codes identified here could account for as much as half of the codes used throughout the entire portion of the planet where one would even imagine finding good weather. Thus, it appeared that this look into climate codes was not necessarily helping much, in the effort to home in on the best weather locations.

Moreover, while some such codes applied only to smallish areas, others were quite expansive. For instance, in North America, the Cfa type covered a zone from San Antonio TX to Lexington KY to much of southern Florida, and the BSk climate region stretched from central Mexico to Canada. The BWh code covered perhaps a third of Africa and at least that much of Australia. For such reasons, it appeared that the zone of ideal weather would be limited, not merely by climate type, but also by temperature and other weather elements.

Many websites offer lists of cities that enjoy what is supposedly the best weather in the U.S. Examples include AreaVibes, MarketWatch, AARP, US News, Suntan.com, and CurrentResults. Several of those lists are particularly oriented toward retirees, and that is interesting: it could be that retirees have more freedom to choose their location according to weather, or it could be that weather is the most compelling thing for many retirees. Or both. Some of those lists arrive at strange conclusions. For example:

AreaVibes suggests that Corpus Christi, TX is one of the ten best weather cities in the America. In support of this view, AreaVibes says, “Average temperatures in the summer hang around the 80s.” That writeup links to a separate webpage clarifying that the August average of 84F results from daytime highs ranging up to 104F. These data are perhaps better examined with the aid of the Outflux tool, which reports that, in most months, Corpus Christi is not only hotter but also windier and more humid than New Orleans — and both are on the FarmersAlmanac list of cities with the worst summer weather.
CurrentResults states that, in its view, Charleston, West Virginia is a runner-up for states with best weather, with a climate that make it a good place to live. But Outflux indicates that Charleston WV has wintertime average low temperatures below 32F, with more precipitation than Seattle, in most months of the year, and nearly as many cloudy days.
MarketWatch identifies Carson City, NV as one of the nation’s best weather cities. But Norton arrives at a very different conclusion. Norton rates locations according to the number of days per year when the mean temperature is between 55F and 75F, the minimum and maximum temperatures are between 45F and 85F, and there is no significant precipitation or snow depth. By those standards, Carson City is in, or close to, one of the worst weather areas in the country. Incidentally, as of 2014, Carson City had experienced a more extreme rise of temperature due to climate change than any other city in the U.S.

The makers of the lists of best-weather cities do tend to agree on a short list of similar locations, notably including cities in southern California (e.g., San Diego, San Luis Obispo), and Florida (e.g., Key West, Sarasota, Tampa), with occasional endorsements of New Mexico (esp. Taos, Las Cruces, and Santa Fe) and Oregon (esp. Bend and Medford), as well as others, almost entirely in the South and West (e.g., Prescott, AZ; Sequim, WA; Lake Charles, LA; St. Marys, GA; Asheville, NC; Grand Junction, CO; St. George, UT; Fort Worth, TX).

Temperature Research

For insights into temperature, I returned to the search cited earlier in this post. There, I discovered a number of interesting research reports:

Steiger et al. (2016) found that, during a seven-day summer trip to the Alps in southern Germany, survey respondents reported an ideal temperature range of 70-77F, whereas beachgoers preferred the range of 77-90F. While Alps tourists who were older or who engaged in sporting activities (e.g., hiking) preferred cooler temperatures within this (70-77F) range, families with children preferred warmer temperatures.
Edwards et al. (2015) found that children studied over a period of five years, starting at age 3, were most likely to be active at temperatures around 65F. For every 10 degrees above 65F, their moderate to vigorous physical activity was reduced by five minutes per day. For every 10 degrees below 65F, such activity was reduced by 17 minutes. Increasing precipitation and wind speed also yielded reductions in such activity.
Hewer et al. (2016) found that, among park visitors, visitation declined when maximum temperatures exceeded 91F in peak seasons (84F in shoulder seasons), and also when temperatures dropped below 52F. They suggested that beach tourists may be influenced especially by daily temperature, while park visitors may be more influenced by multiday averages.
In an analysis of data taken from Americans aged 15 or older in the American Time Use Survey, Zivin and Neidell (2014) found that individuals engaged in an average of about 44 minutes of outdoor leisure per day. They averaged 37 fewer minutes of such leisure on days where the maximum temperature was 25F than on days where the maximum was 76F or higher. Daily indoor leisure increased sharply when maximum outdoor temperatures exceeded 95F. Workers in industries with high exposure to climate allocated less time to labor as maximum temperatures exceeded 85F. At 100F, they allocated about one hour less to labor per day than when the maximum was 76-80F. People who were not employed, presumably enjoying more freedom to choose how they would spend their time, began reducing their outdoor leisure when maximum temperatures exceeded 90F. With maximums over 100F, they averaged 22 fewer minutes of outdoor leisure per day compared to days when the maximum was 76-80F. It also appears, however, that they became somewhat acclimated: on days with a maximum temperature over 100F, they spent 30 more minutes outside in August than in June.
Konnikova (2013) cited Connolly (2013) for the finding that temperatures above 90F have a worse impact on happiness than does being widowed or divorced, and also cited other studies arguing that the optimal temperature is either 72F or 81F (humidity not specified). Livestrong contended that the optimal exercise temperature is 68-72F, but Matz (2013) said the best Olympic marathon times came at about 50F (my own preferred temperature for running distances of greater than one mile) and Bruning (2015) cited research indicating that optimal marathon results call for temperatures as low as 45F (see also Quora). Seppanen et al. (2006) found 72F optimal for office productivity, but Lang (2004) pointed to 77F for the fastest and most accurate keyboard typing.
In the hot and dry climate of Athens (Greece), Tseliou et al. (2013) found that questionnaire respondents felt thermal comfort at outdoor temperatures between 63F and 70F in winter, and between 79F and 90F in summer.
In several laboratory experiments involving college students in rooms heated or cooled to various temperatures, Huang et al. (2013) found that warm temperatures increased participants’ tendency to agree with others’ views, seemingly increasing one’s feeling of closeness with others.
NOAA suggested that the ideal was a “mild” temperature range of 64-86F.
Another search led to a suggestion that indoor thermostats be set to 68F in winter and 78F in summer as acceptable compromises between comfort and energy savings.

These observations seemed to call for a revision of Norton‘s U.S. map, in which a location got, in effect, one point for each day that its mean temperature was between 55F and 75F, and its minimum and maximum temperatures were between 45F and 85F, with no significant precipitation or snow depth. The revision would operate on demerits: no demerits for an effective (i.e., heat index adjusted) average temperature of 65-80F; one demerit per degree per day, for each of the first 10 degrees above or below that range; and two demerits per degree per day, for each of the next 10 degrees beyond that. Thus 65-80F would be the ideal range, 55-64F and 81-90F would be the secondary ranges, and 45-54F and 91-100F would be the marginal ranges. According to the foregoing sources, people would still be relatively active outdoors in the secondary ranges, but outdoor activity would drop off sharply in the marginal ranges, and even more sharply beyond those.

Developing Temperature Criteria

An earlier section identified a number of cities with desirable weather, inside and beyond the U.S. For present purposes, those cities were Albuquerque, Honolulu, Loja, Long Beach, Malaga, Nice, Phoenix, Port Elizabeth, Sacramento, San Diego, San Francisco, Sao Paulo, and Sydney. Remember that some of these were named as ideal by some of the 11 sources listed in the earlier section, while others (i.e., Albuquerque, Phoenix, Sacramento, San Francisco) were my own additions to the list, based on their locations within Köppen climate zones that were supposedly ideal. Although I would be referring to them all collectively as “perfect weather cities,” some — most frequently, those that I had added to the list — had various weather imperfections. Those imperfections would tend to emerge in the following discussion.

At this point, I examined temperature data, in the downloaded WWR table, for those 13 perfect weather cities. (Data for others I would have liked to examine, such as Las Palmas, Kunming, and Medellin, were incomplete or not included in the WWR table.) (Later I would discover the Kunming data were there, but my search failed because the table misspelled it as Kumming.) Those data were expressed in Celsius rather than Fahrenheit, so I would have to provide conversions as needed.

The downloaded temperature data led me to formulate certain criteria related to mean monthly temperatures. Except as otherwise indicated, the approach I took was to look for the least extreme value that would exclude one city as outlier, under each criterion. That value would constitute the limit on perfect weather, for purposes of that criterion. Then, for each criterion, I also looked for a more ideal value, satisfied by just barely a majority (i.e., seven, or 54%) of these 13 cities. The following bullet points describe the reasoning and values that resulted from this method of identifying various characteristics of ideal temperature.

Summer: Maximum Desirable Mean Daily High Temperature. It occurred to me that, actually, people don’t think the weather in the desert is perfect. In the warm season, they turn into snowbirds and head north, or they go inside and enjoy their air conditioning during the daytime. What most people like about places like Phoenix is especially that they aren’t very cold in winter. I saw that the mean daily maximum temperature of nearly 41C (106F) in Phoenix, in the hottest month of the year, was much higher than the maximums in the other “perfect weather” cities. Sacramento, second-highest on the list, maxed out at 33.5C (92F). It seemed reasonable to chose, as a tolerable maximum average summer temperature, the lowest value that would exclude only Phoenix: 34C (93F). Having done that, the more exacting threshold, met by 54% of the cities, was 29C (84F) — and that value, by the way, was also consistent with the foregoing research.
Summer: Minimum Desirable Mean Daily High Temperature. The other question, regarding high temperatures, was how hot it needs to get, to give a person an adequate sense of having had at least one month of summer. The tolerable minimum, excluding only one ideal weather city, was 23C (73F), and the city excluded was San Francisco, whose warmest monthly average was just slightly below that. Weatherbase, evidently using somewhat different data, said that the average high temperature in SF peaked at 70F in September. Either way, apparently SF hardly ever reached 90. This did not seem to be the kind of place where one could reliably lounge around all day in shorts and a T-shirt in summertime. The more ideal minimum average daily high temperature, for the hottest month of the year, met by just over half of the cities, was 28C (82F).
Difference Between Mean Daily High and Low Temperatures. I considered establishing one or more criteria to identify cities with too much, or too little, variation between daytime and nighttime temperatures. For instance, Accuweather said that Sacramento’s summer temperatures in 2016 ranged from daytime highs of anywhere from the low 80s to around 104F, to nighttime lows in the upper 50s. It developed, however, that that range was much less extreme in winter. It seemed that people did not generally prefer temperature extremes; but when dealing with summer heat, perhaps that kind of extreme cooling was welcome. I decided, on balance, that the evidence was not clear enough, from these cities and from my experience of living in different places, to establish a criterion on this point. It seemed that careful attention to high and low temperatures would be sufficient to address most people’s temperature-related priorities.
Summer: Maximum Desirable Mean Daily Low Temperature. Suppose it is midnight in August, and you want to sleep. You could turn on the air conditioning, except you happen to be tenting with your kids or grandkids (or, different scenario, you are trying to save on utility expenses). To help you sleep, what should the temperature be? In Phoenix in August, it would average about 28C (82F) — and, actually, it might not drop to that level until around 5 AM. Wrong answer! It seemed safe to say that, for any month of the year, most people would prefer an average daily low temperature of less than that Phoenix level of 82F. In fact, none of the other perfect weather cities had a mean daily minimum temperature exceeding 24C (75F) in any month. The majority-minority dividing line was somewhat lower: seven (54%) of these 13 perfect weather cities had no month in which the mean daily minimum temperature exceeded 19.3C (67F).
Summer: Minimum Desirable Mean Daily Low Temperature. What if you aren’t ready to go into the tent: what if you want to sit around the campfire, telling stories until the wee wee hours, or you want the water not to be absolutely frigid for swimming tomorrow? The concern here was that the warmest month(s) of the year should meet a certain minimum temperature. I decided not to develop both minimal and ideal criteria for this concern, so as not to give this item too much weight: people did not generally seem to worry that their summer nights would not be hot enough. On the belief that summer should not be chilly, however, I did determine that a bare majority (54%) of these 13 perfect weather cities had at least one month whose mean daily minimum temperature was at least 19C (66F).
Winter: Minimum Desirable Mean Daily Low Temperature. We had the undeniable fact that perfect weather cities would not be heavy on winter. The outlier, here, was Albuquerque, with winter monthly average low temperatures as cold as -3.2C (26F); none of the others had any months averaging below 3C (37F). I was surprised that so many of these perfect weather cities would have relatively cool average daily low temperatures: a majority had at least one month whose average daily lows were less than 7.8C (46F). But then I realized that, as long as the days were warm, it might not matter — it might even be preferable, for purposes of variety — to have cool nights for a part of the year, on the assumption that cool nights would not discourage most outdoor activities. I decided to keep the minimum low temperature criterion (i.e., 3C), but not to add a criterion focused on the daily minimum temperature of the majority of cities.
Winter: Minimum Desirable Mean Daily High Temperature. Winter was fine, as long as it did not mean cold daytime temperatures. A person could stand only so much of that. Among these perfect weather cities, the outlier was, again, Albuquerque: it was the only city on the list whose lowest daily average high temperature, for any month of the year, was less than 12C (54F). A majority of these cities had no months whose high temperature averages were less than 18C (64F).
Winter: Maximum Desirable Mean Daily High Temperature. In summer, as shown above, it was preferable to specify limits in both directions: summer should not be too hot, but it should also not be too cool. Similarly, a climate scheme that reversed the priorities followed in this post — one that valued seasonal variety over temperatures warm enough to get people outside — might insure that winter was not too cold, but also that it was not too warm. But should this post approve any criteria that would favor winter? In practice, this was a question of how to view Honolulu, as it was the only one of these cities that did not have any months with a mean daily high of less than 23C (73F). When I was going through the various articles identifying perfect weather cities, I saw that Honolulu — indeed, all locations in Hawaii combined — were not endorsed as consistently as some others, including San Diego, Malaga, and Sydney. These others did have a recognizable cool season. Based on that (and with the benefit of my years in hot cities, where I grew familiar with what people in such places liked and disliked about their weather), I concluded that, as long as we had multiple criteria penalizing temperatures that grew too cold, we should have a criterion favoring a mild cool season. A majority of these cities had at least one month in which the average maximum temperature was no higher than 19C (66F). But then, as I continued with the following paragraphs, I decided not to use this criterion, as it was essentially repeated in a better form in the season-length criteria (below).
Winter: Maximum Desirable Mean Daily Low Temperature. For purposes of this post, the concept of protecting a cool season did not extend to the point of insuring that a city would experience at least one month of truly bitter winter.
Seasonality. Except for Honolulu, the perfect weather cities endorsed by the travel writers (as distinct from Phoenix and the others that I had added to the list) all had warm and cool seasons. Moreover, their mean high temperatures in the three warmest months of the year all fell within a certain, relatively narrow temperature band, and so did the mean low temperatures in the three coolest months of the year. For the warm-month highs, the monthly means ranged from 22.7C to 29.7C; and for the cool-month lows, the monthly means ranged from 6.1C to 12.5C.

Altogether, then, observation of the perfect weather cities had led to a total of 12 temperature-related criteria, applicable in every case to the monthly means of daily high or low temperatures. I verified that none of these criteria were duplicative or mutually contradictory — imposing the same limiting values and identifying the same (or exactly opposite sets of) preferred cities. To sum up, those criteria were as follows:

Highest high not above 34C: 12 of 13 cities (92%)
Highest high not above 29C: 7 of 13 cities (54%)
Highest high not below 23C: 12 of 13 cities (92%)
Highest high not below 28C: 7 of 13 cities (54%)
Highest low not above 24C: 12 of 13 cities (92%)
Highest low not above 19.3C: 7 of 13 cities (54%)
Highest low not below 19C: 7 of 13 cities (54%)
Lowest low not below 3C: 12 of 13 cities (92%)
Lowest high not below 12C: 12 of 13 cities (92%)
Lowest high not below 18C: 7 of 13 cities (54%)
Mean of high temperatures in 3 warmest months between 22.7C and 29.7C: 8 of 13 cities (62%)
Mean of low temperatures in 3 coolest months between 6.1C and 12.5C: 10 of 13 cities (77%)

Testing and Applying Temperature Criteria:
129 Cities with Perfect Temperatures

I compared the 13 perfect cities in terms of their overall scores on those 12 criteria. San Diego and Sao Paulo scored highest, with 11 out of 12 possible points. In San Diego, the missing point was due to the fact that no month came near to the average high of 28C (82F). In Sao Paulo, a point was lost because the low temperature in the warmest month was very slightly higher than the 19.3C threshold. By these 12 criteria, although Sao Paulo had not been so consistently mentioned by the travel writers, its weather was actually slightly superior to that of universally praised San Diego.

By these 12 criteria, Port Elizabeth, Loja, and Long Beach tied for third place, with 10 points out of a possible 12. Malaga, Nice, and Sydney tied for sixth place, with 9 out of 12. Honolulu got 8, due mostly to its overly warm nights and otherwise erring on the warm side, compared to these other perfect weather cities. Sacramento, Phoenix, and San Francisco all got 7. Albuquerque got 5. While this distribution of points did not entirely endorse the travel writers’ priorities, it did promote the cities chosen by the travel writers over the several cities that I had chosen merely for their location within a certain Köppen climate zone. Incidentally, I noticed that Sydney would have scored higher, but for its habit of recurrently landing very slightly on the wrong side of the criteria. Preliminarily, within the limited testing ground of these 13 cities, it seemed that the 12 criteria might be sufficient to distinguish cities whose temperature patterns were especially desirable. Or at least, as in Sydney’s case, to distinguish regions: for instance, Adelaide (in Sydney’s region) would score 10.

I applied the 12 criteria to the list of 1,759 cities in the WWR 1991-2000 database. I started by deleting cities from that list that lacked data for one or more months of the year, and also those for which there were duplicate entries. That left 1,528 cities. When I applied these criteria to those cities, the results were encouraging: I had a list of 129 cities with scores of 9 or higher — which, in theory, would qualify them to be included in a global list of cities with perfect temperatures. The list was generally believable. Those cities, and their scores, were as follows:

I used the same data to produce country-level scores. These scores were the averages of the scores of the cities included in the database for these countries. If other cities had been included in the database, the overall country averages might have changed. The accuracy of the average would depend upon how well the country’s climate zones were reflected by the included cities. Hence, the ranking would be less meaningful for larger countries. As shown above, a low score for a country would not mean that it had no perfect-temperature cities. In short, the list should be construed as interesting, not authoritative:

At any rate, now I had a temperature database that I could cross-reference against other databases to assess the weather desirability of a given location.

Humidity and Heat Index

According to Wikipedia, people become uncomfortable in high humidity, in warm temperatures, because humid air is less receptive to the evaporation of sweat, by which the human body cools itself; people also become uncomfortable in low humidity, because it dries skin and exposed inner (notably nasal) tissues. Another Wikipedia page observed that the heat index, representing individuals’ subjective reports of perceived temperature similarities in humid and dry conditions, could significantly understate the perceived heat experienced by a person at a particular warm temperature in the shade, if s/he moved into the sunshine, or became physically active (e.g., exercising, working), or was not wearing a hat, or was in a physical state not tested (e.g., menopause, pregnancy, obesity, drug or alcohol effects). The heat index calculation evidently included attention to some such factors.

Wikipedia offered three formulas for calculation of the heat index in degrees Fahrenheit. The first formula was the one NOAA used to calculate its heat index chart, but I liked the second one for its coverage of lower humidities and temperatures, as in this extension of the NOAA chart:

By any of Wikipedia’s three formulas, calculation of the heat index required temperature and relative humidity readings. I assumed that heat index calculations would produce heat index tables, but apparently that was backwards: it seems the original researcher (Steadman, 1979) had produced a table, and the formulas were designed to recreate its values. An alternate approach, the humidex, used dew point rather than relative humidity to produce values that, unlike the heat index, did not attempt a statement in degrees. For example, a humidex value of 40 to 45 would mean “great discomfort; avoid exertion” without any reference to any particular temperature. (I saw that NOAA’s Heat Index Calculator allowed use of dewpoint to calculate heat index values. Unfortunately, my searches did not lead directly to either a statement of the relevant formula or to dewpoint data for cities worldwide.) Accuweather’s proprietary Real Feel approach claimed to achieve superiority over competing measures, such as the heat index, by including other factors (e.g., sunshine, wind speed) but, as with Intellicast and other sources, it did not appear that data were downloadable in table or spreadsheet form, so as to facilitate comparisons among multiple cities; rather, the user would have to look up each city individually, and would get only current data, not monthly or other averages.

A companion webpage suggested setting indoor humidity at 30-40% in winter and 60% in summer, with about 50% being optimal as a single year-round setting. The U.S. National Weather Service (NWS) offered a Heat Index Chart suggesting that 80F at 40% relative humidity (RH) really feels like 80F, but that the relationship breaks down as temperature and/or humidity rise above those levels. NWS also offered a calculator, apparently valid at or above 80F/40%RH, to compute heat index based on temperature and RH. (NWS also offered a wind chill chart and calculator, effective only at or below 40F.)

Other Materials and Links

See also: yearly sum of direct irradiance, Meteo Info maps, heat index search, charts, and calculator, wind chill search and calculator, wind speed and direction maps (NullSchool, Hint.FM), average monthly temperatures (U.S.), daily air temperatures & heat index data (U.S.), NOAA Climate at a Glance data (U.S.)

General-purpose weather sources: Weather.com, WeatherUnderground, WeatherSpark, Accuweather, Weather & Climate, Weatherbase

How Much Hotter Is Your City Now (U.S.)?

Humidity & Sunshine Across the U.S.

How Many Pleasant Days Per Year in Your City (U.S.)?

Cities with Highest Average Humidity (U.S.)

My downloaded NOAA weather data for the 1990s gave me mean monthly temperatures (i.e., averaging highs and lows). These did not seem to add much to what I had already figured out. The NOAA data also included total monthly precipitation (in millimeters). This, too, did not seem as informative as other weather and climate items that I would ideally be able to obtain, such as information on humidity, cloud cover, and extreme weather events — although I doubted I would find good worldwide city-by-city data on all those items. Precipitation data could be informative on drought and flooding; but as I had seen in the case of snow, sometimes cities that get more of it are more prepared for it than cities that get less.

Generally, these “perfect weather” cities enjoyed above-average sunshine. The average of all monthly values (i.e., average for January, for February, etc.), for these cities, was 262 hours of sunshine per month, as compared to only 195 hours for the full set of 1,698 locations. (Note that some countries had many entries in this set, and some had none, so that average of 195 would not necessarily be the average of all cities or countries worldwide.) There was considerable variation among these “perfect weather” cities. But aside from the very sunny Phoenix (323 hours per month, on average) and Sacramento (301), the others (i.e., Sydney, Nice, Malaga, Albuquerque, L.A., San Diego, and San Francisco) had between 208 and 285 hours of sunshine per month, for an overall annual average of 248.

At this point, things got a bit tricky. I saw that Fort Wayne IN, near where I grew up, averaged 232 hours of sunshine per month. In my opinion, Fort Wayne was too cloudy. More precisely, I felt the real problem was the winters: they weren’t just cold, or gray; they were long and cold and gray. Not to deny the occasional February day of brilliant sunshine; but from November through February, Fort Wayne saw, on average, only 135 hours of sunshine per month — as compared against an average of about 334 from May through August. Even more so, Portland OR: its overall average of 195 was exactly the average for these 1,698 weather stations, but Portland saw an average of only 76 hours of sunshine per month, from November through January.

These bits of information inspired me to calculate moving four-month averages for each place throughout the year (e.g., Jan-Feb-Mar-Apr, Feb-Mar-Apr-May). It seemed to me that four months of relatively unchanging cloudiness was long enough to feel really long. Compared against very gray winter locations (e.g., northern Russia) offering as low as zero hours of sunshine per month for an entire four-month period, the “perfect weather” cities were remarkably pleasant, with at least 152 average hours of sunshine per month in even their grayest four-month periods. And yet that 152 (for Nice) was uncomfortably close to Fort Wayne’s 135. Similarly, New York had 154 and, while I hadn’t spent a lot of time staring at the sky in Manhattan, I couldn’t say northeastern New Jersey was exactly the Sunshine Capital. Likewise Boston, also with 154 average hours of sunshine per month during its grayest four-month period, had seemed pretty gray, when I lived there. On the other hand, I’d found that San Antonio’s grayest four-month average of 163 hours of sunshine per month was fine. No doubt these subjective assessments were influenced by temperature and also, perhaps, by life conditions. After looking through the list and considering other locations I was familiar with, I decided to draw the line at 160 hours of sunshine, during the grayest four-month period. As with other criteria developed in this post, this wouldn’t necessarily be a veto on any particular city; it would just be a factor helping to identify ideal places.

There was also the possibility of an excessively long summer grind. Upon re-sorting the list to look at four-month periods of maximum sunshine, I saw that Sacramento was the sunniest place on Earth, according to these data, with an average of 418 hours of sunshine per month, from May through August, followed by Fresno with 413 hours. Phoenix averaged 387 hours per month in that same period. Albuquerque, another arid location, had 339. Otherwise, the highest moving four-month averages for the “perfect weather” cities ranged from averages of 229 to 320 hours per month in any consecutive four-month period.

At what point would abundant sunshine become unpleasant? A search turned up references to places that various people had described as oppressively sunny. These included various arid locations (ranging from Colorado to Israel), coastal California, Houston (274), Melbourne (240), and even Paris in the summer (220). Most likely, these descriptions were affected by (a) the nonstop year-round prevalence of sunshine (in e.g., desert locations) and/or (b) the heat. For instance, I didn’t recall feeling that Fort Wayne’s average of 329 hours of sunshine per summer month were too much, even as a kid out in that sunshine, except when I was too hot: the sun was such a welcome contrast against the long winter.

In my review of the list of 1,698 places, sorted by their four-month periods of maximum average hours of sunlight per month, I was tempted to choose 310 hours as the very rough dividing line between enough and too much sunshine. Above 310 hours, the places I recognized or knew personally were increasingly those of drier and hotter lands (in e.g., Spain, the western U.S., Egypt), where a person could readily consider the sunshine oppressive at some times of year. I knew that the sun in Denver (309 hours) could be brutal — at least until you step into the shade and freeze. But a 310-hour limit would penalize half of the remaining “perfect weather” cities (i.e., after discounting Phoenix et al.): Malaga (320 hours), Los Angeles (317), and San Francisco (315). A 320-hour limit would keep those cities — and, above 320 hours, I didn’t see too many places renowned for their pleasant weather. The others on the perfect weather list tended to be well below that 320-hour limit: Nice (303), San Diego (274), and Sydney (229).

Climate Change

The foregoing analyses were predicated on climate stability. I would have preferred to use more recent data, but in a sense it didn’t matter: any forecast would be susceptible to revision due to climate change. That would be true, not only of temperatures, but also of precipitation, storm patterns, sunshine, and other elements of climate. Sci-fi is full of alternate scenarios for the future of our weather — none, perhaps, more famous than Blade Runner, with its visualization of a rainy Los Angeles.

At this writing, global warming denial was no longer credible; anyone could see spring arriving earlier, the polar ice melting, glaciers retreating, temperatures generally warmer than average (via e.g., NOAA, or the graphs offered in the Monthly view at AccuWeather.com), and heat records being set virtually every year (via e.g., 1 2 3 4 great visualizations). But it was not clear how bad things might become, or how soon.

NASA (2015) offered detailed climate change projections, but (a) these data appeared to be in a format that was not familiar to me and (b) the climate scientists seemed to be continually revising their forecasts in light of new information. For instance, Yale Environment 360 (Jones, 2016) suggested that climate change could dramatically impair the Atlantic Ocean circulation pattern that prevents Europe, at its relatively high latitude, from going into “a deep chill.” At the same time, the Times offered maps and graphs suggesting that, by the year 2100, various cities in the U.S. would experience at least three times as many days over 95F as they experienced a century earlier (i.e., in 1991-2010).

Not many expat retirees were going to worry about the weather in 2100. It presently seemed that climate change could significantly alter the weather in various places over a period as short as the next 30 years. In some cases, the changes could be so dramatic as to compel a relocation. But it was not sure what the expat could do about that, other than perhaps favor larger countries where there would be more internal relocation options. In short, climate change seemed to be a factor worth keeping in mind, perhaps allowing a bit of slippage in the temperature and other constraints discussed above; but it was not clear how climate change could be explicitly factored into the discussion of preferred weather.