Explanation of the Data We Generated for The Wall Street Journal

We recently generated some data for a Wall Street Journal article entitled "In the Forecast: A Better Way to Measure the Temperature". The article appeared in the April 7–8, 2012 edition of the US Wall Street Journal, on page A2 (the page immediately after the front cover). You can read it online and, if you click the "Interactive Graphics" tab at the top of the online version, you can see the full set of data that we generated for The WSJ.

We've received a number of questions about how exactly that data was calculated, so we thought we'd explain it here for all to see.

The aim of the analysis was to rank the 2012 list of global cities in terms of temperature comfort (how comfortable their climates are with regard to temperature).

Other comparisons with a similar goal have typically been based on average temperature data. But average temperature data is deeply flawed because it averages out the temperature variations that have a major impact on the comfort experienced over any given time period.

For example, consider a day with an average temperature of 72°F. That sounds pretty pleasant. But the averaged temperature hides all the detail that went into it. That day could, for example, have seen temperatures ranging between 50°F (which most would consider to be a little on the chilly side) and 92°F (which most would consider to be too hot). There may have been precious little time in the "Goldilocks Zone" (as the Wall Street Journal put it) where the temperature was just right.

When temperatures are averaged over weeks, months, or years, the variations are lost further. All in all, comparisons based on average temperatures are crude approximations at best.

To get a better idea of a location's temperature climate, you can study detailed charts of temperature variations over the course of a year or so. But that is a pretty involved process – not much good for comparing a number of locations at once.

Degree days offer an accurate, easy way to compare different temperature climates for comfort. Or at least they do when the degree days are calculated properly by taking into account the temperature variations within each day (which of course our system does).

The assumptions and methodology used for the WSJ data

For the Wall Street Journal analysis, we started with the following assumptions:

• Temperatures between 60° and 80° Fahrenheit are comfortable. The WSJ aptly termed this the "Goldilocks Zone".
• Comfort decreases when the temperature strays outside of the Goldilocks Zone.
• The further the temperature strays outside the Goldilocks Zone, and the longer it strays outside for, the less comfortable the climate overall.

For each of the 66 global cities, we calculated:

• The heating degree days (HDD) with a base temperature of 60°F. This value represents the time and extent to which the temperature dropped below 60°F (i.e. was colder than ideal).
• The cooling degree days (CDD) with a base temperature of 80°F. This value represents the time and extent to which the temperature rose above 80°F (i.e. was hotter than ideal).
• The sum of the HDD and CDD above. This total value indicates the total time and extent to which the temperature strayed outside of the 60° to 80°F range.

We generated the figures above over the 1-year and 5-year periods immediately preceding the date of publication. We summed together the daily degree-day values over those periods to get the 1-year and 5-year totals. Using the 5-year average figures that our system generates automatically would have worked just as well.

To make the figures more accessible to a broader audience, The WSJ decided to divide the sum of the HDD and CDD by the number of days in the period of consideration, to give the total number of HDD and CDD per day over that period. Degree days per day is effectively just degrees (as the day parts cancel out), so the figure left after the division is essentially the average number of degrees that the temperature deviates from the 60°–80°F range. If you're already familiar with degree days, you might find the figures easier to understand if left in degree-day form.

We also ran a similar analysis, using 70°F as the definition of "comfortable". For these figures we calculated both the HDD and the CDD with a base temperature of 70°F. These figures are interesting for comparison, but we think the 60° to 80°F figures are much more useful given that comfort is a range (especially when you allow for variations in sunlight, wind and humidity), and the fact that there is nowhere on earth that doesn't have diurnal (day/night) temperature variations which, we think, makes it rather pointless to define a single outside temperature as ideal.

To see the results, visit the online version of the WSJ article and click the "Interactive Graphics" tab. Then click the "Average number of degrees temperature deviates from 60°–80° range in the last five years" column header to rank the locations using that measure of temperature comfort.

Given that 60°F is a fairly typical base temperature for building heating, and 80°F is a fairly typical base temperature for building cooling, this should also approximately sort the locations in order of heating/cooling energy consumption. (We say "approximately" because, across the locations, buildings would vary considerably in terms of their usage and thermal properties. But it would be a fair comparison to make if you could imagine placing the same building in each location to cancel out those differences.)

Trying a similar analysis yourself

Different people have different ideas of what constitutes a comfortable temperature range. If you wanted, you could run an analysis like this with your own definition of the Goldilocks Zone. If you wanted to consider hot temperatures to be worse than cold temperatures, or vice versa, you could multiply the CDD or HDD by factors to weight them differently before summing the weighted values together. This would be a little like how heating and cooling systems typically have different efficiencies and so the simple sum of HDD and CDD is rarely ideal as an indication of total energy consumption.

If you wanted to run a similar analysis on a large scale (across lots of locations), we would recommend that you automate the process to make it faster and less prone to manual error. You could use our desktop app to quickly assemble HDD and CDD for all your locations into a single spreadsheet, then use Excel formulas for the rest of the analysis. Or you could get a software developer to write a little program to assemble the data automatically using the Degree Days.net API.

For a small number of locations you could easily use our free website to find the necessary weather stations and download HDD and CDD for each before combining the data together in a spreadsheet.

In fact, why not give it a go right now for the location of your building, to see how its climate compares with that of the global cities used for the WSJ analysis?

Specify and Generate Your Degree Days Now