photo of sky at start of az=90 observation (.jpeg)


plots:

normalized caldeflection vs elevation (.ps) (.pdf)
tsys vs elevation, the fits (using Tatm=270K) , and the fit residuals (.ps) (.pdf)
   the fits using Tatm=290K (.ps)  (.pdf)

a closer looks at the Tatm contribution to Tsys as well as the bump around el=60 deg (.ps)  (.pdf)
Tsys fit values vs frequency (.ps)  (.pdf)
fitting different elevation ranges to the tsys vs el atm contribution (.ps) (.pdf)

Intro

    Tsys vs elevation was measured on 20sep22. The measurements  consisted of:

• The telescope was driven between 7 and 87 degrees elevation at .5 deg/sec
• az=0 drive up and then drive down
• az=90 drive up and then drive down.
• datataking started at 14:30 and ended around 15:00
  
• While driving the telescope the mock spectrometers took data:
• 7 172 MHz bands were taken. cfr: 8219.00 8363.00 8505.00 8647.00 8789.00 8931.00 9073.00 MHz,
• 512 channel across each 172 MHz band was used.
  
• the   hardware cal was run during the experiment.
• spectra were sampled at 2 milliseconds.
  

Processing the data:

• the idl routine masgetscanhwc()  was used to input the data.
  
• The total power in degK is computed for the calon,caloff data for each 40 millisecond cal cycle.
• the TsysCalOff is then used for the rest of the processing.
• the idl routine p12mfittsys()  was used to fit to tsys vs elevation. The  fit was:
• Tsys=Tconst + Tatm*(1-exp(-tau*secza)) + Tcmb*exp(-tau*secza)
• Fit for Tconst and tau
• set Tatm =270 K, Tcmb=2.725k

Plotting the results:

    The calcycle is 20ms calOn, 20ms caloff. Since the telescope was moving at .5 deg/sec ,  TsysCalOn will differ slightly from TsysCalOff. The
cal difference will then include this small change.
The first plots looked at whether the cal deflection was biased by the elevation motion.

The plots show the  normalized caldeflection vs elevation (.ps) (.pdf)

• The 8219 Mhz band was used for the plots.
• The up and down scans at az= 0 were used.
• Page 1:
  
• Top: Normalized Tsys vs elevation:
• Each curve was normalized by their median value.
  
• black: PolA 0 to 90 degrees moving up
• green:PolB 0 to 90 degrees moving up
• red    :PolA  90 to 0 degrees moving down
• blue   :PolB  90 to 0 degrees moving down.
• middle: Tsys rate of change while moving.
• green and black are moving up in el (so tsys is decreasing)
• red and blue are moving down in el (so tsys in increasing).
• there is very little change in tsys above about 30 deg el
  
• Bottom: CalDeflection vs elevation (normalized to median value)
• let eps be the change in Tsys during the 40ms  cal cycle.
• CalDeflectionM= (calDeflection +tsys) - (tsys + eps) = calDeflection - eps
• When moving up eps is < 0 .
• When moving down  eps is >0.
• The blue,red moving down shows a constant ramp in CalDeflection over the entire el range
•  But eps was > 0 only for el < 30 deg. So the slope we see is not from the motion of the telescope.
• Page 2: the cal deflection change repeats for the motion at az=0 and az=90
• the change in cal deflection could just be and electronic gain variation... but
• Top: PolA cal Deflection for up,down at the two azimuth positions
• Bottom: PolB cal deflection for up,down at the two azimuth positions.
• The change in cal deflection:
• repeats for scans going up in elevation at different azimuths.
  
• repeats for scan going down in elevation at different azimuths.
• So the change in cal deflection is not an electronic gain change (unless the electronic gain depends on the direction of el motion).
• The change in value is 1% of the cal value.
  
• if it is a level change rather than a gain change then it is about 32K*.01=.3K
• I'm not sure what could be causing this?

The 2nd set of plots shows tsys vs elevation, the fits (using Tatm=270K) , and the fit residuals (.ps) (.pdf)

• Each page shows a different freq band.
• All 4 scan (az=0 up,down, az=90 up,down)  are over plotted in different colors.
• Top: Tsys vs elevation and fits for polA and polB
• middle: fit residuals  (data-fit) for polA
• bottom: fit residual (data - fit ) for polB
• Looking at the residuals there is a small bump around el=60 deg for all bands.
• The fits are too high at low elevation and a bit low at high elevation.
• I repeated the fits using Tatm=290K (.ps)  (.pdf)
• this  gives the same fit residuals.
• Change Tatm only changes Tconst and tau some

The 3rd plot takes a closer looks at the Tatm contribution to Tsys as well as the bump around el=60 deg (.ps)  (.pdf)

• Top Atmospheric contribution to Tsys.
• The Tconstant parameter from the fits was removed from each Tsys data set.
• Az=0 up direction was used. All 7 freq bands polA,B are over plotted.
• All freq bands and pols overlay one another.
• At el=90 deg the Tsys contributions is about 6.3K ( this is atmospheric and any scattered radiation).
• Bottom: A blowup of the atmosphere Tsys contribution 50 to 90 deg el.
• The slope changes between elevation of 60 and 70 deg.
  

The 4th set of plots has Tsys fit values vs frequency (.ps)  (.pdf)

• Top: Tsys vs frequency for elevation = 60 deg.
• the small error bars are the fit sigmas.
• Middle: the Tconst fit value vs frequency
• this includes Trcvr and any other constant temperature values.
• The 4 scans overlay each other pretty well.
  
• The + are just to show where the data was measured. The coef sigmas were much smaller than this
• Bottom: The opacity (tau) from the fits vs frequency
• Black in polA, red is polB
• The value  is .015 to .017.
• Not sure what the dip  at 8931 means (may be weak rfi).

processing: x101/220920/tipper.pro

---

Tsys from atm and scattered radiation

    The atmospheric contribution to tsys has a sec(za) dependence with  za. Scattered radiation should also have an dependence on za (or el)

•  You might think that the largest contribution from scattered radiation occurs at el =90 deg
• The secondary over illuminates the primary and sees the ground.
• As you move to lower elevation some of this spillover will start to see the cooler sky.
• This el dependence would be opposite to the atm contribution (largest at low elevation).
• I tried fitting different elevation ranges to the atm contribution and then looked to see if the residuals told us anything about the scattered radiation.

The plots show the results from fitting different elevation ranges to the tsys vs el atm contribution (.ps) (.pdf)

• I ended up fitting 6 elevation ranges:

• el range (deg)	color	notes
  6 to 88	red	all the data
  6 to 33	green	bottom 1/3
  6 to 48	blue	bottom 1/2
  33 to 60	light blue	middle 1/3
  48 to 88	light brown	upper 1/2
  60 to 88	purple	upper 1/3

• Top: Tsys vs el fits
• Black is the data
• Bottom: fit residuals (data -fit)
  
• If you don't fit the bottom 33deg, then the fit values at low el are too large (but up to 8 degk)
• If you include the bottom 33deg  in the fit then the maximum residual is about 1degK.
  
• This may not be saying much, since the strongest curvature is at the lower elevations.
• It doesn't look like you can separate out the Tatm from the scattered radiation by playing with the elevation ranges.l

processing:220920/chkfit.pro

---

SUMMARY

• elevation strips 7 to 87 degrees and back were done at az=0 and 90 degrees.
• Tsys was fit to elevation using Tsys=Tconst+Tatm*(1-exp(-tau*secza)) + Tcmb*exp(-tau*secza)
• Tcmb=2.725
• Tatm was set to 270K (290K was also tried)
• there was good agreement in the fits at different azimuths.
• If the cloud conditions had changed, then it wouldn't have been constant.
• Tconst and tau(opacity) were fit for.
• tau was about .015.
• If we changed Tatm then Tconst and tau would vary.
• At el=90deg there was about 6.3K from the atmospheric fit.
• the slope of tsys vs el changed around el = 60 to 70 deg.
  
• The cal deflection varied over a strip
• this variation differed going up and down in elevation
• but it repeated at different azimuths (so it probably wasn't a gain variation with temperature).
• We looked at whether you could separate out the Tsys from the atmosphere  and from scattered radiation by playing with the elevation range using in the fitting.
• nothing obvious  popped up.