This pattern is in fact the signature effect of a phenomenon that has been termed the ‘urban heat island’, or UHI, as we’ll call it throughout this article. Simply put, a UHI is a bubble of air over an urban area that’s hotter than the air in the surrounding rural areas. This effect has been identified scientifically for many years, and it’s certainly obvious anecdotally to anyone who’s watched or read a weather forecast in Hong Kong – how many times are the forecast nighttime lows something like ‘17 degrees in the urban areas, and a few degrees lower in the New Territories’?
But if you're happy with a quick overview, then we can just say that a UHI forms when the sun heats up man-made structures such as roads, masonry, buildings, and so on. This heat builds up during the day, and it's then gradually released after the sun goes down. This keeps the surface temperature from dropping as much overnight as it would in an area covered with natural vegetation.
Some scientists who have studied urban heat islands also believe that heat-generating machinery, e.g. vehicles and big air-conditioning units, contribute to higher temperatures in urban areas. It's hard to judge how big an effect this would be, but if it's a real factor, Lord knows Hong Kong is a place it would make plenty of impact.
So the obvious question for us now is, how much of the warming in Hong Kong is due to the heat island effect? And then, if we can identify that number, how much background, or ‘global’, warming has really taken place in Hong Kong?
Fortunately, the busy staff of the HKO have adressed these very questions. A 2006 paper titled 'On Climate Changes Brought About by Urban Living', by C Y Lam, pretty much has the goods we need. He conducts a simple comparison that's typical of attempts to estimate the magnitude of individual UHI effects: he compares the temperatures at the urban station in question, in this case the HKO headquarters (which we'll call 'HKO-HQ' for short) itself, with a nearby location that's still rural. In fact, Lam chooses two such locations, i.e. Lau Fau Shan in the NW New Territories, and Ta Kwu Ling in the far northern NT, very near the border with mainland China. Lam limits his comparisons to the years 1989-2005, which are the years for which data from Ta Kwu Ling are available.
That period is of course also the time in which the HKO-HQ has recorded its sharpest increases in average yearly temperatures: Lam notes that the HKO-HQ's average yearly temperature has risen at a pace of 0.37 degrees C per decade during this time. This is a very steep rise indeed.
So how does that compare with the two non-urban stations? In Lau Fau Shan, from 1989-2005, the temperature increased by 0.25 degrees per decade. That's still very fast. For Ta Kwu Ling, however, the increase is significantly lower, i.e. 0.08 degrees/decade.
What does this tell us about the UHI effect at the HKO-HQ? Keeping the math simple, we can conclude that the UHI effect at the HKO-HQ in recent years is somewhere in the range of 0.12-0.29 degrees/decade: that is, 0.37 - 0.25 = 0.12 for the HKO-HQ/Lau Fau Shan comparison, and 0.37 - 0.08 = 0.29 for the HKO-HQ/Ta Kwu Ling comparison. Lam himself, interestingly, does not derive even this simple range of difference in his paper, nor does he speculate on how great the UHI effect at the HKO-HQ might be. He simply notes:
'That the urban area has been warming up much more rapidly than the "countryside" is thus evident'.
Now the range I came up with range is very large: which end of it is likely to be most accurate? I suspect the UHI effect is at the higher end. For one thing, Lau Fau Shan is not exactly a rural area; it's fairly built-up. Several researchers from the HKO note this themselves a later paper, i.e. 2007's 'Temperature Projections in Hong Kong Based on IPCC Fourth Assessment Report':
At LFS (i.e. Lau Fau Shan), changes in the environment were observed near the site in the past few years. The grassland in the vicinity of LFS was converted into concrete surface for stacking up cargo containers. Also, there had been continuous new town developments at Tin Shui Wai which is about 1 km away.
If you'd like at look at the area, I entered the coordinates for the Lau Fau Shan station provided by the HKO, and you can see where it's situated on Google maps:
The Ta Kwu Ling station, on the other hand, seems to be near a village, but it's situated in somewhat more open territory:
The HKO confirm TKL's better fit as a rural comparison to the HKO-HQ in the 2007 temperature projections paper:
There is no significant change in the immediate environment for TKL although the rapid urban development in the nearby city of Shenzhen cannot be completely ignored.So we've made some progress. It's obvious that the urban areas of Hong Kong comprise a textbook urban heat island. We now also have at least a broad sense of how much effect this may have had on the HKO-HQ's sharply rising temperatures. The HKO, to its credit, does not shy away from discussing the UHI effect, nor does it try to understate its impact on Hong Kong. C Y Lam concludes his paper with a rundown of some the harmful outcomes of a hotter city -- e.g. more very hot nights, stress on people who can't afford air conditioning, etc, -- but he in no way tries to imply this nighttime heating is purely the result of global warming.
Still, it would be nice if we could find way to nail down the scope of the UHI effect in Hong Kong more definitively.
The HKO make their best attempt at estimating the UHI at the HKO-HQ in the 2007 temperatures projection paper I've cited above. The authors first examine the same comparison Lam made in his paper, i.e. between the HKO-HQ and Ta Kwu Ling. They then run a linear regression on the differences recorded (i.e. they attempt a far more sophisticated statistical analysis of the yearly data than the seat-of-the-pants business I did). They reject the trend that they identify as statistically insignificant, however, because the data sample is small (it's not even two decades' worth of readings) and because of large yearly variations.
So the HKO writers then adopt what they call a 'differencing period' approach, which I'm not going recount in detail (it's explained on pages 4-5 of the 2007 paper if you're interested). In brief, they assign the temperatures at the HKO-HQ from 1885-1902 as their 'base', since at that time they claim the HKO-HQ was in an essentially rural setting, i.e. paddyfields, then compare it to the current gap in temperatures derived from the HKO-HQ/TWL comparison. They come up with an 'urbanization' rate of 0.08 degrees/decade, i.e. this means that every decade since 1902 the temperature at the HKO-HQ has risen 0.08 degrees strictly because of urbanization, for a total of 0.81 degrees' worth of UHI.
The HKO authors then compare this number to studies of UHI done in other places: for example, a study in London identified an effect of 0.11 degrees/decade; a study in Beijing 0.16, a study of large cities in China 0.05, and so on. Given this range, their number looks reasonable.We can now do a very simple comparison to estimate just how much of the warming in Hong Kong is likely due to the UHI effect. We've got our 0.08/decade number, and we can quite easily determine the overall warming the HKO-HQ has recorded over the years. The 2004 'Climate Change in Hong Kong' paper I referenced in my previous article states it clearly on p. 8: temperatures at the HKO-HQ have increased by 0.12 degree/decade over the period their records cover.
So, 0.12 - 0.08 = 0.04.
In other words, by this estimate, Hong Kong has been warming by 0.04 degrees Celsius per decade when we eliminate the UHI effect. Put another way, the UHI effect seems to account for about 2/3 of the warming Hong Kong's urban areas have experienced.