Reports
Contents
| Title: | Sonic Comparison | ||||||||||||
| Date: | 2021-02-10 - 2021-03-02 | ||||||||||||
| Data File: | TZ_SonicCompare_20210302.csv | ||||||||||||
| Refers to: | TZ,170523,153805 | ||||||||||||
|
On the Tonzi Tower from 2021-02-10 to 2021-03-02 Gill sonic sn 153805 which had been partially submerged when the Bouldin Corn tower fell over was compared with the existing Gill sonic sn 170523. The existing system (170523) recorded data at 20Hz the USB drive in the Licor Li-7550 box as GHG files. These 20Hz GHG files were then processed by Matlab script to produce 30min averages. The comparison system (153805) was measured by a CR1000 at 1Hz and saved as 30min averages. The headings of the two sonic anemometers were different so wind direction will not be compared here but could be done later if necessary. Instead wind speed calculated as the SQRT(U*U + V*V) is compared. There is a similar sonic comparison in the report 47: Tonzi Tower Eddy System Compare. Figure 1. Time series of horizontal and vertical wind speed. In the early morning hours of 02-15 the existing tower sonic (170523) had a strange excursion in all channels. But other than that the horizontal wind is very tight. Both vertical winds are bouncing around zero as expected and match in magnitude.
Figure 2. Comparison of 30min horizontal wind speed. Very tight match here with slope near one, almost zero offset and very little scatter. Cleaning out the 25 points from the excursion makes little difference. Report 47 had similar regression statistics: m=1.0079, b=0.0027096, r-sq=0.9913, n=596
Figure 3. Comparison of 30min vertical wind speed. Because these values bounce around zero the regression here is not expected to be good. With some creative zooming here you can get the axes to be roughly equal and see that this scatter is more of a round blob. The sonic SOS is also compared against one another and against an HMP155 that was mounted on the tower at the same time. Sonic anemometers are notoriously bad at measuring absolute air temperature. Sonic temperature is proportional to the square of any error in the distance between sonic transducers - a change of 200 micrometers in sonic path length at room temperature yields a 1°C change in measured sonic temperature. It would probably be better to look at the SOS (or Ts) variance than absolute temperature but that will have to wait until both units can be recorded at the same frequency. Figure 4. Time series of SOS. The two sonic anemometers follow each other well but there is a pretty consistent offset of about 2m/s with 153805 reading higher. Again the Tower sonic 170523 shows the excursion in the early morning hours of 02-15.
Figure 5. SOS vs SOS. The scatter in this plot doesn't look much worse than other inter-sonic SOS comparisons. This is a relatively small range though of only about 10m/s. The comparison in Report 47 had worse slope and offset but less noise though a somewhat larger range of comparison: m=1.1429, b=-48.146, r-sq=0.9918, n=596
Figure 6. Ts vs HMP155. Sonic temperature (sos^2/403 - 273.15) - see the discussion above of sonic temperature. The fit from the Tower sonic versus the HMP45 previously was better (Report 47: m=1.0042, b=-1.7150, r-sq=0.9960, n=596). However this was over a 25 degree range instead of the 15degree range here. Taking a look the Ts variance would better evaluate this comparison. |
|||||||||||||
| |