Degree days are degree days, right? Does it really matter where you get them from...?
Well, yes it does! We don't like to brag, but we feel it's important for newcomers to energy-data analysis to understand that there are compelling reasons why 5,000+ energy professionals download data from Degree Days.net each month (and often much more frequently):
We don't use a crude approximation method based on daily average temperatures. We calculate degree days properly, using the full detail of the temperature records available.
It's a lot more work, but it's worth it. Hot days and cold nights don't average out in energy consumption, and they shouldn't average out in degree days either.
Buildings vary significantly in terms of their thermal properties and usage, and consequently their base temperatures. With Degree Days.net you can stick with the "defaults" (e.g. 65°F, 18°C, 15.5°C), or, for improved accuracy, you can use the real base temperatures of the various buildings you analyze.
You usually want degree days to match your metered periods of energy consumption. Degree Days.net covers most common use cases by offering daily, weekly, and monthly data (as well as multi-year averages). And you can always sum daily values together to match irregular consumption records (this is a perfectly valid thing to do).
Virtually all real-world weather stations have some irregular, missing, and erroneous temperature readings. Even the high-quality airport weather stations, like the NWS stations in our system.
Degree Days.net has been carefully programmed to handle these problems, and it highlights the extent to which data is affected by interpolations. We don't try to hide the fact that data quality varies, and our "% estimated" figures and star ratings are very useful when choosing weather stations.
Degree Days.net provides degree days for locations worldwide, using the same calculation processes for all of them. Using our system, data from country A is directly comparable with data from country B. This is not the case if you're getting data from multiple different sources using different approximation methods.
With our desktop app and the API that powers it (and many other external software systems), it's easy to pull large quantities of data out of our system. This is critically important for energy managers in large multi-site organizations and energy professionals that manage consumption for a large number of clients.
Point 1 above deserves some further explanation. It's hard to explain it succinctly, but it's important.
Degree days are defined as the integral of the differences between the outside air temperature and the base temperature, over time. If the outside air temperature drops below the heating base temperature, the building needs heating (i.e. non-zero HDD); if it rises above the cooling base temperature, the building needs cooling (i.e. non-zero CDD).
To calculate degree days properly requires detailed temperature records. But detailed temperature records haven't always been available. And even when they were available, in the days before computers it wasn't really feasible to turn them into accurate degree days on a large scale. So people made do with approximation methods. The simplest of all being the Mean Temperature Method (e.g. 65 - ((max + min) / 2)).
To most mainstream weather-data providers, degree days are just a simple add-on. Take the average daily temperature and subtract it from 65°F - job done - let's get back to the forecasting...
But that's simply not good enough for accurate energy-data analysis! The problems with approximations like this are easy to understand:
Consider a building with a cooling base temperature of 65°F. If the outside air temperature rises above 65, the building needs cooling to stop it getting too hot. Now consider a common climate: hot days and cooler nights. In the day, the temperature rises to around 80; at night it drops to around 50. (max + min) / 2 gives us an approximate average temperature of 65, so, according to the mean temperature method, the degree days are zero and the building supposedly needs no cooling...
Yes, it's pretty obvious that's not the case. Or at least it is to anyone with a building, a thermometer, and a vague awareness of their HVAC system. Hot days and cold nights don't cancel each other out, same as hot summers and cold winters don't cancel each other out. Real-world thermostatically-controlled buildings simply don't work like that.
Unfortunately the mean temperature method assumes that they do. And, despite the fact that detailed temperature readings and powerful computers to process them are now both readily available, many degree-day-calculation systems are still using that outdated approximation method. Either because it's the easiest option for them, or because they simply don't know any better.
We calculate degree days using methods that take into account the temperature variations within each day. For most of the good weather stations in our system, we use the full detail of the temperature records available - 60-minute readings, 30-minute readings, 20-minute readings, sometimes even readings taken every couple of minutes. Whatever the weather station records, we use.
In many climates the differences are only small. But the calculation method becomes much more important in climates where the outside air temperature frequently straddles the base temperature.
Our calculation process is many times more complicated and computationally expensive than it would be if we'd settled for a simple approximation method. But, as a specialist provider of degree days, we feel it's important to calculate the data as accurately as we can. And data from Degree Days.net is a better indicator of real-world energy consumption as a result.