I use Cyclemeter to track rides on my iPhone 6, then when the ride is over, I send the info to Strava. When I look at the ride in Strava the calculated Calories burned for a 17 mile 90 minute ride is ~500 while Cyclemeter says ~1100 (My weight is 225lb). I do not have a heart rate or cadence monitor, so the only info being used is the GPS (distance and elevation data) and barometer(I assume) and then my weight. What factors could account for such a big difference? Do the 2 services use vastly different data from maps? Which one should I trust?
As mattnz alluded to it can be notoriously difficult to estimate calories burned, even with a heart rate monitor. As a bit of history, for the longest time it was generally believed that Polar heart rate monitors were the most accurate, because Polar essentially purchased the rights to the best proprietary algorithm available (going by memory, I will need to find the reference for that statement).
The take home is that even if you have a high quality heart rate monitor you may still not be getting an accurate estimate of actual calories burnt as it can vary brand to brand.
Without a heart rate monitor (as you are doing now) you should see even less accuracy. That said, of the two I would hazard an educated guess that Strava would be more accurate due to the depth of data they have at their finger tips. Now none of us know what analyses they are doing behind the scenes, but if they are leveraging any of that immense data archive, they should be able to provide better estimates, assuming they have a high quality statistician working (or consulting with) the company.
Hmm... I should submit a cv...
RANT - Do not read if you do not want to be offended.
A lot of calorie burn questions popping up at the moment. The problem is biometrics has become the latest fashion accessory, and everyone (manufacturers and users) are clambering on the band wagon without stopping to think what they want this data for. Manufacturers want to sell gadgets and make money, users are buying into the marketing hype. Ask yourself why do you want to know how many calories you burnt exercising? So you can justify the cream doughnut you are about to stuff in your mouth, or because you are an ultra endurance athlete struggling to main high enough calorie intake, or something in the middle?
Calorie burn can only be accurately assessed in a lab by measuring your O2 intake and output and calculating how much O2 you burnt. A very close approximation is a power meter, especially if used in conjunction with the results from a lab test. The reality is we are all born different, and just like car engines, some of us are more efficient than others, some idle less efficiently (Basal rate) and some produce power more efficiently. To cap it off, how efficient we are depends on the food (car analogy - fuel) you put in.
A usable, but not great approximation for calorie burn can be achieved by using a heart rate monitor, your weight and height, and the results from max heart rate test ideally including a V02Max test, and tables collated by measuring lots of people of different shapes and sizes and presuming you are average (Keeping in mind the average person is not average).
Given this complexity exceeds the capacity of Marketing people to comprehend, they simplify it to a point its useless. If you buy a device and it claims to tell you how many calories you burnt, you can believe it - but if you continue to believe after a few minutes of research, I bet you also believe the earth is flat and man never landed on the moon. If you don't believe me, ask yourself how the above apps account for wind in their calorie burn calculation. Do they ask for you bike weight for hill climb calculations?
However, the equipment is useful (if you are a bore at parties) - if gives you a number you can brag about.
Some apps will use the GPS altitude information - and the errors on that can be quite large at ground level with clutter (basically all the satellites you can see are above you which doesn't help). Here's an example: Until a few days ago I was using cardiotrainer and the last 3 times I ran the same 10km loop I got 59m, 22m and 0m total climb. I'm confident this just gets altitude from the GPS.
Other apps/services will have access to terrain mapping which will pick up a different set of altitude changes (maybe more, maybe less depending on the dataset).
Some won't use any altitude data at all but I would assume that Strava does and that's your lower figure so unless you're riding downhill all the way we can assume that cyclemeter does too.
Then you get in to how they smooth the track data in the horizontal (more noticeable running, but you get cheated on tight turns).
A big difference will come from what it assumes about drag (and your own efficiency but lets stick to the mechanical stuff for now). If 1 app is calibrated for a road racer, tucked right down, lycra, 23mm rock-hard tyres and smooth tarmac it will assume vastly less power than an app assuming a city bike with wide tyres, some tread,an upright posture and a scarf trailing behind. Let's take a guess and assume that Strava likes road bikes.
What assumptions do they make on the downhills? We don't know, but you might or might be pedalling hard or resting, and the speed alone is insufficient to know which given the other variables.
If we assume (erroneously) that the apps I've used in the past are spot on, then with me being a little lighter than you the truth lies somewhere in between, probably closer to the higher figure (I was getting just under 500cals for 10 miles with some hills in 40-45 minutes, across a few apps).
So what use is it? Well, it lets you compare routes (do you want a workout or an efficient commute, are you trying to train and explore the area at the same time). If you're training seriously your efficiency and your basal metabolic rate will be affected, and both need a lab to measure them (therefore values are assumed in the apps, another source of difference).
You could if you wanted (and I've done this so I know the data is out there) take your speed profile and plug it into an equation for the required power (most sources are for road racers but there are different drag coefficients out there). Then you'd need to use simple physics on your total climb to work out how much work you did, convert this to calories and take into account human efficiency, values for which vary hugely. This would be a laborious way to pick which app you're going to trust.