M2EA thermal accuracy in freezing ambient temperature.

[MikeFox]

Greetings all Mavic pilots - and especially Enterprise Advanced pilots.

I have a request for comments regarding the thermal sensor accuracy when flying at around freezing ambient temperature. I'm getting wildly erroneous temperature readings. My readings are FAR too low!

When measuring miscellaneous materials (different emissivity) I get 10 - 20 degrees C below what's expected. My sensor seems to be very fond of the specific temperature -20.1 degrees which I often get when pointing the camera towards a clear (featureless) sky at night.
However, my temperature readings for indoor temperatures (~20 degrees C) seems to be spot-on.

I've tried both low and high gain setting with the same results, but skewed with a factor of 10 degrees. Letting the drone acclimatize thermally before flight, or even before booting, doesn't make any difference.

This makes me wonder if there's a thermal calibration routine bug, adjusting too much when going towards (and below) freezing weather conditions. I've flown down to -20 degrees C and readings which I expected to be around zero turned out to be -20.1C which to my best guess is the lowest possible temperature presented.
I've tried to change the temp scale to both Farenheit and Kelvin to eliminate a possible conversion error but the results seems to be consequently linearly lower to the actual (presumed) temperature.

As I stated, I've measured a lot of different scenes with different materials in them (concrete , soil, vegetation, metal roof, glass, etcetera) - using MAX, MIN and Average measurement, but no readings are even CLOSE to what I'm expecting.

Reading my own face for example, in freezing weather conditions, reads around a few degrees C when it's supposed to read ~36C. Which it does when testing indoors.

Now I have to set my alarm temperature at 0 or 1 degree C to discriminate anything living while it's freezing cold outside and I can't do any sensible readings during inspection.

Is there anyone else out there who's flying in freezing weather, doing SAR for example? What's your experienced thermal accuracy?

If this is a common occurrence we need to get it fixed. So far DJI hasn't replied to my question in their forum.

Is there some calibration I can do that I'm missing?

 

[jaja6009]

You seem to have a strong grasp on thermography, but here are my comments.
There may also be a problem with your camera, but here is my take on some of your issues.

When reading your face, the most reliable place to get a temperature is the tear duct. To do this you would have to have the camera inches away, as a good point of reference is at least 3x3 pixels on target to get a good quantitative reading. The tear duct is too small for this task.

Also, the cold weather would affect your temperature as well as humidity, wind and TReflc, distance, ambient temperature. Any wind on such cold days will have a cooling effect on objects. Make sure you have a proper viewing angle of perpendicular to max 30 degrees.

I also get a similar temperature of the cold sky as you get.

Our drone lacks proper thermography settings while in flight. You can take an image and use the DJI Thermal Analysis Tool and then change the parameters (Emissivity, Humidity, distance, TRFlec but lacking ambient temperature) and you may get better results.

Check your temperatures with another thermal camera, thermometer or thermocouple.

I do fly SAR for Public Safety and at any useful altitude a person will be no where near the 98.6 degrees body temperature for reasons stated above. When flying high there are simply not enough pixels on target along with the environmental issues listed above.

I have tested my M2EA against a Seek, FLIR handheld, and two FLIR cell phone thermals and all of the thermals listed performed very similar in their radiometric data.

I will do some testing soon to gather some data to share with you.