Quite a while ago some guys on the old forum helped me make a thermal anemometer for arduino. It's been a while since I did anything with this project, but since I have received a few messages on this topic I wrote a fairly short blog post on my anemometer. You can find it here:
I hope it is of use to someone.
p.s. Should I add this to the playground "environmental" section? I have no idea how...
I just saw this article, hence I'm posting quite late. But I thought it might be useful to point out that a sensor like this is expected to have significant errors at low flow or "still" air due to the convection flow from the self-heated sensor. The device that was used for calibration of this project specifies a lower limit of 0.125 m/s for this reason. This is the main reason for the significant non-linearity shown at the low end of the calibration curve and input data values below a certain level should be discounted as suspect and not "linearized".
And for what it is worth, this same principle can be used to measure vacuum levels - here the assumption is use in a non-flowing gas and the voltage/heating required to maintain the sensor at a set level is proportional to conduction of heat from the sensor which is proportional to gas pressure.