The backlight disrupts the temperature reading

Thanks for the video and your post! A change from 0°C to -6°C is equivalent to about a 10° change in Fahrenheit (from 32°F to ~21°F).

This seems like strange behavior, but I'd be surprised if this is related specifically to the backlight being on or off - the backlight should not generate much (if any) measurable heat, nor would I expect the Edge's internal thermometer to go down.

A few photos displaying how the Edge is mounted would also be helpful.

What you are displaying here is the Edge's built-in temperature sensor data. In very hot or very cold riding conditions, you may find the ambient temperature from the Weather widget more useful.

I have been in contact with Garmin support and they have started investigating the issue. The bug has not been reproduced on their devices, but I have now sent them additional data. My device seems to be faulty. I will update this thread with more information once there is something new to report.

And apparently, the cold had nothing to do with the issue. The device behaves the same at room temperature. If anyone knowledgeable about electronics has any guesses about what might be causing this, feel free to add them to this thread.

The temperature sensor is internal and will be affected by many factors including display backlight use & brightness, charging, sunlight, mounting position, wind, etc.

I am following our investigation and curious how you typically make use of the temperature data field.

If your phone is typically connected while riding, I would suggest testing local weather data in place of device temperature.
Garmin Connect displays local weather data for synced rides, and the Estimated Body Temperature Data Field can be very useful depending on what your training goals are.

The response from Garmin support was disappointing because it revealed that they didn’t quite grasp the problem. The response mentions temperature rising, even though the issue is about the temperature value dropping. If the temperature were indeed rising, I could accept their explanation, but since the opposite is happening, it feels like they didn’t fully understand the issue. So, I’ll continue using a faulty device. 

Here’s the translation of their response:

Thank you for your response and the additional information you provided. I discussed the issue further with our product development team, and according to them, the problem in this case is due to the device’s design. Its temperature sensor reacts strongly to the device’s internal temperature, which may increase, for example, as a result of changes in display settings or situations where the device’s battery heats up. If a separate screen protector has been installed on the device, it is also possible that this prevents heat from dissipating from the display, which could then cause the device’s internal temperature to rise.

I can see clearly in your video and activity data that device temperature goes down (temporarily) when the touchscreen is selected and backlight is turned on. The mounting position, use of screen protector, battery charge, backlight settings, and external factors are all expected to impact your device's temp data.

I again suggest that local ambient weather data or estimated body temperature may be more useful depending on what your training goals are.

Edit: double post

Of course, it’s good that there are other options to monitor temperatures, but it’s a bit puzzling that no one else has noticed this issue.

This suggests that it might be a specific characteristic of this device rather than a general problem. If the discrepancy were just a degree or two, I might believe the explanations, but a rapid deviation of over five degrees °C clearly points to something else. To me, this appears either as a faulty device or a programming error.

Anyone can still test this themselves: Set the backlight to maximum (max only because it makes the test easier, more light, more error), ensure the battery level is preferably below 50-60%, tap the screen, and observe what happens. For me, the readings start rapidly dropping within about 5-10 seconds, even when the device is placed on a table (place doesn't matter). And as soon as the backlight turns off, the same happens in the upward direction. I'm pretty sure this isn't happening for anybody else. 

820/530 never had anything like this. Have to say, I'm a bit disappointed. 

EDIT: 

And in the name of science, I even sacrificed the screen protector and removed it. It had no effect on the issue.

I have also been able to reproduce temp drops when backlight ON (both indoors & outdoors) for Edge 1040 and have escalated this for further investigation.
It would help us most to know how you typically make use of the device temperature data, otherwise I will update here with any more info.