Sonic Wind Calibration Notes
These are my rough notes. The figures indicated are working figures
(like Figure 9., and
Figure 10. that I used for
my analyses. They do not refer to the figures on these www pages.
<![CDATA[
WINDS
Only u and v components examined, no attempt to correct w component
We assume u,v,w are alligned perfectly.
Small variations from this shouldn;t matter too much
Usually not enough info from one period to get slope info so looking at all periods
Slope (i.e. differences = f(u) could be caused by uneven orientation of sonics
or more likely tower-induced flow distortions
To check for this I plotted the residuals of the fit vs rel wd.
U1,V1 refer to lower hazer values, whatever the actual height.
U5,V5 refer to upper hazer values, whatever the actual height.
U2-4,V2-4 reprent the fixed level measurements.
Added information based on shebacal2 program
looked at all data - scaled between levels - restricted to Z/L <.07 both levels
This is referred to as "cal2" data...
Bottom hazer vs level 2 (figures 7)
U-COMPONENT (figures 7)
Period 1 Level 2 slope explained by tower blocking lev 2 but not hazer at rwd ~ 120
Period 5 Level 2 higher for same reason
Period 6 level 2 was higher than hazer but this can be explained by rwd ~210
which brings wind through tower to hazer but clear to level 2 making
Because U is negative at this rwd this blocking causes U1 - U2 to be > 0
After eliminating questionable rel wd periods there still appears to be a bias
of about 0.35 m/s i.e. level 2 is biased lower (not magnitude but actual)
Cal2 data verifies above bias = 0.35; no slope
Conclusions: Add 0.35 m/s to all level U2 values
V-COMPONENT (figure 10)
Compare very well throughout entire experiment except a few outliers period 1
These explained by rwd ~125 - tower blocks hazer but not level 2 (same as u)
Period 6 Level 2 higher same reason as u-component
Cal2 very similiar, does seem to have bias ~ -.1
Conclusion: Bottom hazer, level 2 virtually identical, no correction required
Bottom hazer vs level 3 (figures 8)
U-COMPONENT (figures 11)
Two periods obvious from cal2 comparisons before and after JD 400
This is when the sonic was removed. New heads installed?
Period before JD 400
U1-U3 ~ 0.2 + 0(U1)
Period before JD 400
High Period 3 values due to ~rwd which causes tower to block hazer but not lev 3
After accounting for rwd effects there still appears to be Bias = 0.32 Slope =
-.098 i.e. U1-U3 ~ 0.40 - .098(U1)
Conclusion: U1-U3 = 0.2 + 0(U3) before JD 400
U1-U3 = 0.4 - .098(U3) after JD 400
V-COMPONENT (Figure 11)
Bias (at V=0) appears to be zero.
Obvious tower effects during period 1 and some of periods 2 and 3
There does appear to be a slope effect so that V1-V3 ~ -.12(V1)
However Period 5 doesn't fit this pattern. Fo this one V1-V3 ~ -.05(V1)
cal2 suggests slope= -.0230 before JD 400 and -.0665 after JD 400
But period 1 calibration seems to fit after JD 400 line
while Period 5 seems to fit before period.
In checking cal2 by month I noted the followig pattern
which was consistent with the calibration periods (JD> 400):
Strong trend of V1-V3 = 0 - .12(V1) V1< 5 just like calibrations
V1-V3 = 0 - .04(V1) V2> 5 e.g. period 5
So trend line flattens out at V1>5
Conclusions will be based on mean slope including all data
There will always be a lot of uncertainty in this calibration
Conclusion: V3 = 0 - 0.0230(V3) before JD 400
V3 = 0 - 0.0665(V3) after JD 400
Bottom hazer vs level 4 (figures 9)
U-COMPONENT
Obvious tower effect outliers for Some Period 1 points
The rest follow a very definite pattern U1-U4 = 0.29 - .063(U1)
Cal2 has slightly less bias and flatter slope but not too different
Will go with calibration period data due to very nice fit.
Conclusion: Add 0.29 - .063(U4) to all U4
V-COMPONENT (Figure 12)
Large scatter within and between periods
Again obvious tower effects for some Period 1 points
After ignoring these Best fit is V1-V4 = -.11 - .050(V1)
Cal2 verifies mostly but slope correction seems to large at V4>4
Conclusion: Apply -.11 - .050(V4)
Bottom hazer vs top hazer (1st figure 2)
U-COMPONENT
No apparent rel WD effects as expected since no wind from 180-270 range
Fairly consistent trend for all periods: U1-U5 = 0.16 - 0.090 (U1)
Unfortunately poor verification from cal2 data.
Cal2 has near zero bias and slope. But when looking at all data (not just neutral)
this seems to be closer to calibration fit: U1-U5 ~ .16 slope ~.05
Decided to go with less slope to better match Cal2 and entire data set
Too much scatter to make calibration after sonic board repaired JD 608
Conclusion: U1-U5 = 0.16 - .05(U1)
V-COMPONENT (2nd figure 2)
Consistent trend except for period 1: V1 - V5 = -0.16 - 0.034(V1)
Period 1: -0.2 - 0.037
Period 1 had higher V values
General trend verified by cal2 including higher wind anomaly
Went with fit to lower winds
Large scatter but same approx calibration after JD 608
Conclusion: Apply -0.16 - 0.034(V5)
Top hazer vs level 4
U-COMPONENT
Moderate scatter and no negative components
Vague fit: U4-U5 = -0.3 - 0.016(U5)
Cal2 gives same bias but has slope = 0
Bias consistent with U1,U5 and U1,U4 comparisons slope a little large
After sonic board was fixed and re-installed on JD 608 installed, there
was a clear change in calibration.
After JD 608 U4-U5 = -0.2 - 0.055(U5)
V-COMPONENT
Alot of scatter
Vague fit V4-V5 = 0 + 0.03(V5)
Generally verified by cal2 - bias appeared a more positive
Fairly consistent with U1,U5 and U1,U4 comparisons
After JD 608: V4-V5 = 0 + 0(V5) (match well)
]]>
|
|
| Outline Page |
Sonic Wind Page |
|
Last update: 5/4/99
Please send all comments and suggestions to the author, Peter
Guest,