...
Following the SCP project, we tested each sonic in a zero-wind enclosure within the EOL temperature chamber, measuring wind component offsets and sonic temperature errors over the nominal range -30 C to 55 C. The procedure is to warm the chamber up to 55 C and then slowly decrease the temperature linearly to -30 C, followed by slowly increasing the temperature back to 55 C. The zero-wind enclosure holds two sonic heads on their sides (with the v-axis vertical) and electronics, one above the other and separated with a horizontal layer of rigid blue foam. On the cool-down cycle, the atmosphere around the top sonic is unstably stratified (cold enclosure lid above warmer the air) and that around the bottom sonic is stably stratified (cold enclosure bottom below warmer air). The opposite is true during the warm-up cycle. Both the wind component data and the sonic temperature data have significantly less variance when the air is stably stratified, and thus we use data from the bottom sonic during the cool-down cycle and data from the upper sonic during the warm-up cycle.
...