Hi there

I use a Mavic 2 Enterprise Dual and was recently taking some thermal point measurements of a snow covered grass surface. Outside temperature was about 5 degC and the snow was slowly melting. However, the point measurements I took at different points of the snow blanket were about 12 to 14 degC in the app, which seems way too high for what I would expect. The same happened later over an icy lake. I would expect to read around 0 degC on both occasions.

Any idea why this could be? Is there a possibility to recalibrate the IR camera or do I have to let it run for a while before taking measurements?

Any help and suggestions appreciated to get the correct temperature measurements.

Cheers
/Jack

Doesn’t sound that far off.  You’d need to take serious measurements to check the calibration.  

A common phenomenon. The thermal chamber itself is not thermostamilized but is volatile, from there large calibration differences are obtained. It is very important at what angle the measurement is performed, because the ground layers of air fluctuate a lot within tens of centimeters for a few degrees, especially the morning evaporation 1-2 hours after sunrise. Above ice is also important turbulent ground flow and humidity just above the ground. also significant is the height of the measurement, the higher the temperature is measured by 1-2 degrees per 10 m for example. These anomalies are characteristic over meadows with grass, bushes, shade under trees. Do an experiment on a larger area, tilt the thermal imager below 10 degrees and keep the liver at a height of about 1-3 meters. You will see exactly the scales of temperature anomalies in a given area.
It is best to fix the drone at a certain height and turn the camera vertically down. you put a contact thermometer on the measuring point and report the difference as a correction value. Then you perform similar measurements at the places of interest.

Thanks for the great replies and hint. I managed to do a test run over a surface of ice sheets melting in water with varying vertical distances to target (i.e. flight height). There appears to be a trend towards slightly higher observed temperatures with increasing distance up so about 15m after which it flattens out a bit. But the effect is quite small (about 2 degC) compared to the expected offset (about 12 degC), but that agrees well with what you suggested, Syntropix. I attached a plot of the measured temperatures below.

I wasn't able to test the angles yet as it was a quite cold and windy day and we ran out of juice quite quickly. But we do normally only take images straight down at a 90 deg angle.

We'll do another test in the next few days with a thermometer on the ground. In the meantime, does anyone know if there is a setting to adjust the emissivity value for the thermal imaging in the DJI app (iPhone)? A search in the forum yielded no results besides "there should be but there is not". Any chance anyone was successful in this regard?

Thanks a million!

xconsolex Posted at 2-11 09:24
Thanks for the great replies and hint. I managed to do a test run over a surface of ice sheets melting in water with varying vertical distances to target (i.e. flight height). There appears to be a trend towards slightly higher observed temperatures with increasing distance up so about 15m after which it flattens out a bit. But the effect is quite small (about 2 degC) compared to the expected offset (about 12 degC), but that agrees well with what you suggested, Syntropix. I attached a plot of the measured temperatures below.

I wasn't able to test the angles yet as it was a quite cold and windy day and we ran out of juice quite quickly. But we do normally only take images straight down at a 90 deg angle.

I did a similar test and took a few screenshots. Absolutely identical case. I measured with an available thermometer the temperature is actually -2 degrees C at  the ground and has not changed for several days. Look at the thermal ranges shown by the camera.
The measurements were made after nearly 45 minutes of flight.In my opinion, the problem is in the calibration and I have doubts that the DJI deliberately do not correct this problem. After 2 two years of requests, the users have so far not made an rJPG output of a file. I think this is either a marketing strategy or they can't really calibrate the camera.