OASIS98: RADIATION Messages: 11 Entries..

I climbed the tower about 11:00 to check if the Everest had the correct
emissivity reading.  It is set to 0.97, rather than 0.98 that I had expected.
I did this because we've seen about a 3 degree difference in Tsfc calculated
from the Eppleys (using 0.98) and the Everest.  I left the setting at 0.97.

Also, while I was messing around, the yoke shifted on the tower a bit.  
I think I got it back in place and tightened the wing nut, but there is
a possibility that it is now looking at a different patch.

At 09:00 I cleaned the domes of the four component array
and the Q7.
I also checked the level of the leveling platform (Swartzenager).
It was level, N-S, E-W, to ~ 0.2 degrees.

Inspection of the radiation record from the beginning of the program until
July 20, indicates that July 15 was a most nearly perfect day. The
24 hour period: July 14, 18:00 until July 15, 18:00 was used to compare 
radiometer responses.
Before undertaking the analysis, the pyrgeometer corrections were evoked by:
                       dpar(robust=F)

A comparison was made between the three four components system: NCARsum, OKMNsum and CNR1asum.
For these systems the four components sw.in, sw.out, lw.in and lw.out can be 
individually examined as well as algebraically added to form a summation 
or "sum"
The NCARsum and the OKMNsum both use Eppley radiometers, whereas the CNR1asum 
uses Kipp and Zonen radiometers.
Note that the  CNR1asum is designated Rnet.CNR1a, but is, in fact, a sum

sw.in: NCAR and OKMN agree well during the day, with a maximum difference
              OKMNsw.in < NCARsw.in by 20 Wm-2 at noon.
       The CNR1a shows a greater difference with a maximum difference  
              CNR1asw.in < NCARsw.in by 60 Wm-2 at noon
       At night the OKMNsw.in is set, by software, to be 0.0 Wm-2, 
       The NCARsw.in has an instrumentation response of - 4 to -2 Wm-2. 
       The CNR1asw.in also shows this instrumentation response. 
       For the CNR1a the value is -10 Wm-2.
These nighttime negative instruentation responses are due to a real export
of energy by the radiometers. The treatment of the OKMNsw data is probably
reasonable, but masks the problem, which also occurs during the daytime.

sw.out: In the day time the three systems differ over a range of 25 Wm-2 at 
        noon. As the three look down at different surfaces this can be due
        to the albedo differences. Thus the OKMN albedo is greater than that
        for the NCAR site because a different grass cutting method is used
        for the land under the OKMN radiometer stands. There is also more
	surface shading directly under the radiometers on the NCAR stand by
	the CR10 box, boom and 4-component mount.
        At night the OKMNsw.out is set by software to be 0.0 Wm-2. 
        The NCARsw.out and the CNR1asw.out show instrumentation 
        responses of -2 and -5 Wm-2

lw.in:  The OKMNlw.in response is very similar to the NCARlw.in response, both
        in the day and at night, with the exception of the 15 Wm-2 higher
        value for the 3 hour period around noon. This anomoly may be due to
        photodegradation of the vacuum-deposited silicon coating on the
        pyrgeometer dome. 
        At night the CNR1alw.in shows a fairly consistant offset of
              CNR1alw.in > NCARlw.in of 20 to 30 Wm-2
        In the daytime CNR1alw > NCARlw increases with a maximum difference of
        +60 Wm-2 at midday.
	This could be due to an incomplete dome heating correction or even to 
	sensitivity to sw radiation.

lw.out: The response of the three radiometer to the emission of lw radiation
        by the surface is simple.
             OKMNlw.out < NCARlw.out by 5 Wm-2 at night, 
             OKMNlw.out > NCARlw.out by 5 Wm-2 by day
        Similarly
            CNR1alw.out > NCARlw.out by 20 Wm-2 by night
            CNR1alw.out > NCARlw.out by 35 Wm-2 by day
The difference of both the OKMNlw.in and the OKMNlw.out to the NCARlw appear 
to be offsets rather than proportional differences in gain.

Comparison of net radiometers and four components summations. 
The ******sum parameters are generated from the algebraic addition of the
four components. The ******net parameters correspond to the output of integral
net radiometers. Note that although the K&Z four component output is designated
"Rnet.CNR1a", it is a summation product, and here is termed CNR1asum.
Six systems were investigated: 
NCARsum, OKMNsum, NCARQ7, OKMNQ7, CNR1asum and NRLite.
The general agreement was good with the only significant disagreements being
the OKMNQ7 at night and the NRLite at noon. The midday anomoly of the OKMNlw
is obvious.


Responses of the sum/net systems compared to the NCARsum: 
                Night-time                  Noon
OKMNsum:     +10 Wm-2                -25 Wm-2  if noon anomoly removed

NCARQ7:      +10 Wm-2                -15 Wm-2
         
OKMNQ7:      +25 Wm-2                -20 Wm-2

CNR1asum:     +0 Wm-2                -30 Wm-2

NRLite:      +15 Wm-2                -50 Wm-2


Responses of the sum/net systems compared to the CNR1asum:
               Night-time                   Noon
NCARsum        0 Wm-2                +20 Wm-2

OKMNsum      +10 Wm-2                -25 Wm-2

NCARQ7       +10 Wm-2                  0 Wm-2

OKMNQ7       +25 Wm-2                  0 Wm-2

NRLite        -10Wm-2                -40 Wm-2

There is a discrepancy between the responses of the uplooking NCAR lw radiometer
and the uplooking OKMN lw radiometer. This is well illustrared on Jul 3, a
clear, cloud-free day.

The Rlw.in plot shows an earlier and higher peak for the OKMN than for the NCAR.
Note that the Rlw.out plots show agreement between OKMN and NCAR.

The Tdome and Tcase plots showed that the OKMN were 1-2 degreeC warmer than 
NCAR, but no structure which could lead, in compensation correction, or 
dome/case correction to the structural difference showed in Rlw.in.Note that 
the difference of 1-2 degreeC is significant.

The Rpile.in plot shows the strong feature.Several aspects were considered:
  - Ventilation: the OKMN is ventilated differently than the NCAR. However
                 the dome and case temperatures plots do not show the strong
                 feature. 
  - Level: to account for the phase shift a tilt of ~11 degrees would be needed
  - Photodegradation: the Si vacuum deposited coating inside the dome is
                      susceptable to uv degradation. This could allow a non-
                      uniform illumination of the photopile. What history 
                      does this radiometer have?

It was noted that immediatedly after sunset the response of the incoming
and the outgoing short wave Eppley radiometers indicated negative short wave
radiation fluxes.
To test whether this was an offset due to the Campbell electronics, the input
from the two radiometers were sucessively shorted, with a jumper between pin 1
and pin 2. In each case to apparent output decreased to +_ 0.078 Wm-2
This value of ~ +_ 0.1 Wm-2 is the data system uncertainty.

It is surmised that the indicated negative radiation flux is due to thermal
effects within the Eppley radiometers. The inside junction of the thermopile
views the warm inside of the radiometer case while the other junction 
views the inside of the dome. The thermopile is sensitive to long wave
radiation and so responds to the flux of long wave radiation from the warm
interior of the sensor to the cooler dome.

The artifact is most severe for the first hour after sunset when it corresponds
to ~ 5 Wm-2. Then,throughout the rest of the night, the artifact maintains a
value of ~ 2-3 Wm-2.

As the artifact is common to both upward facing and downward facing radiometers,
the effect should almost cancel when the net radiation flux is calculated.

NOTE The K&Z short wave radiometers responds in a similar fashion
We suspect that the OKMN short wave radiometer data is forced, by software, 
to yield 0.00 Wm -2, whenever its output goes negative.

  Tony brought some new fans for the 4 Comp rad system.
The Rsw.out now has a fan.

 The Rlw.out fan failed. At the moment, we do not
have a spare. The Rsw.out fan was moved over to the
Rlw.out location.

19:04 UTC:  Leveled the Eppley SW incoming and LW incoming.

Adusted level of net radiometer.  The bubble was touching the edge
of the bullseye and I rotated the radiometer to the east to center it.

A scatter plot of Rnet (REBS Q7) versus Rsum (4 Eppleys) shows:

Rsum > Rnet 	Rnet > 150 W/m^2
Rsum < Rnet	Rnet < 150 w/m^2

A linear fit of Rsum to Rnet give 

Rsum = 1.11*Rnet - 0.66

but there is noticeable non-linearity in the scatter plot.

Leveled radiometer array using two-axis bubble level; checked okay 
(+- 0.1 degree) with one of the hand-held electronic levels.

Radiometer electronic levels read:

lev.rad.x	1.14 +- 0.03
lev.rad.y	0.23 +- 0.03