formatting link
The idea is
Wait and then measure ADC, Vbe reflecting baseline temperature
pulse M2 on and heat Q1 for a while
Kill M2 and measure ADC again, hot temp
The diference in ADC measurements depends on self-heating of Q1, which is less when there's air flow.
I'd love to model the heating, including time constants, but see no easy way to do that and drive the sim temperature. May as well breadboard.