IR Techniques

I have been exploring the capabilities of the MLX90640 Far Infrared (FIR) imaging sensor as a cloud detector. The MLC90640 measures temperatures over a 110x75 degree piece of the sky via a 32x24 pixel array. That works out to about 5x5 deg pixels but that could be sufficient to recognize clouds. The idea is that the clear sky is very cold compared to clouds or other things in the sky (haze, dust, smoke).

I procured a sensor from Adafruit and interfaced it to a Teensy 4.0 microcontroller which reads out the basic data. This data is then read out by a Raspberry Pi 3B+ via the USB cable that also powers the Teensy (which powers the MLX90640). Various Python programs running in the Raspberry analyze and present the data for interpretation. Among the presentation types are gray scale images indexed from lowest to highest temperature, false color images with colors indexed to actual temperature values, statistical measures including lowest temp, percentile readings (that indicate percentage of sky temps less than a given temperature), etc. It is my hope and plan to relate these to crowd-sourced definitions of "great", "good" and "poor" observing conditions with these judgments ultimately displayed in near real time on the ASEM webpage.

Some early results are shown below.

Color coded images

See color scale below

Update 2021-10-22

More data gives more opportunities to study different sky conditions. A new analysis tool is to just look at the mean temperatures along with the standard deviation of the whole range. Below is the current result from that tool. The day started mostly cloudy but the clouds started breaking up around 11:30. The graphs show that the sky was relatively warm early on but very uniform. At the marked point (11:45) the sky was still mostly warm but there was the variability was much greater. By 20:45 the sky was much cooler but fairly uniform.

At 20:45 lots of high clouds could be seen visually due to the ambient light pollution and a nearly full Moon in the East. Jupiter and Saturn were easily seen as were a few stars (most of the Big Dipper in the NW). With the 32 in telescope, M13 could be seen fairly well. Based upon the naked eye scene, I did not think to try M57 which is some 3 magnitudes fainter than M13 (wish I had). With these conditions (especially the bright Moon), I think most visual observers would stay home - a darker Moon might have tempted them.