Some thoughts after talking w/Tom 12/16/98: 1. Our SSSF decision about purchasing levels for the ATIs is split into two parts: a. Should we have levels (will we use them?) b. If the answer to a. is yes, which sensors should we get? As input to b., we know that 3 of the 6 level sensors were broken (electrolyte gone) when they were converted from box to boom-mounted electronics. Also, one of these broke when being tested by ATI when the last sensor was refurbished. Furthermore, Tom has seen a temperature drift in the FLATLAND data which would cause readings to be erroneous. Thus, I would be tempted not to use the ATI sensors if something more reliable was available. The Applied Geometrics sensors have behaved resonably well, now that we understand their calibration, with the exception of one failure during SHEBA recalled by John. This failure should be followed up. Applied Geo. does make a new version of their sensor which is amenable to boom mounting. Also, there obviously are other sensors on the market. We need to decide from a. what is precision is needed from these sensors. If we only use it to decide that the sensor has been removed, we can make do with a pretty coarse (read cheap) sensor. As for a., there are several possible uses: - setting up the sonic - ensuring that the sonic hasn't moved since set-up - actually applying a coordinate rotation into gravity-oriented coordinates. Tom feels that this last is unlikely, though I had hoped to use this as a verification of our tilt correction approach someday. I (and Tom, too, to some extent), do not trust the data from the levels enough to use them for set-up (especially given the temperature drift noted above). In fact, it is only necessary to get them approximately vertical (assuming that we will apply our aerodynamic tilt correction later). Thus, I would simply level the tower with a bubble level or theodolite and assume that our clamping system is adequate to keep the sonic booms nearly perpendicular to the tower. It seems to me to be poor engineering to "kink" or otherwise misorient a tower to follow the signal from a level which has not been thoroughly tested. The one circumstance in which we might want a level measurement during set-up that I can forsee is in a Cheeka Peak-like setting, in which aerodymanic leveling (fitting the flow to a horizontal plane) cannot be used, it is difficult to shoot the tower or sonic for precise leveling, and the tower boom mounts may be sloppy. However, in many such situations the tower will be sturdy enough to allow us to use a hand level while climbing the tower. Thus, Tom feels that the primary use of levels would be to check for movement during the program. In fact, I rarely do this in data analysis, and instead look at vertical velocities, the tilt plots themselves, and verification of changes using logbook. I would think that a pretty coarse sensor (and possibly only in one angle - pitch) could be used to detect movement of the instrument. I note that SHEBA is a special case, in which we would want to continuously monitor the orientation of the TOWER! (Though in this case, with a wider range than ATI's +/- 3 degrees.) In summary: Use Tom Steve Set-up Use if confident in signal, Do manually when needed. but not needed to great Better to make solid booms. accuracy. Movement Yes. Could be useful. Rotation No. Should check in during a few programs. Conclusion: There is a weak case for having levels, but they are only useful if they are stable with temperature and in time. The ATI sensors do not meet this specification. Accuracy to much better than 1 degree probably isn't needed and maybe only from one axis.