Comparison
of Shortwave Radiation Measurements with the Streamer Radiative
Model
Phil Anderson, 9th October 2009.
Overview
In 2004, measurements of
global, diffuse and reflected shortwave radiation were made within the
Instrumented Clean Air Sector at Halley station (76 S 26 W). These have been compared to the Streamer radiative model for clear sky days.
http://stratus.ssec.wisc.edu/streamer_web/userman/intro.html
Comparison
of diffuse radiation for these days allows the tauhaze
parameter to be fine tuned to give agreement of scattering. This then allows
the pointing error of the global radiation pyranometer
to be estimated.
Similar
radiation data are available for 2003, but errors in the shade ring position
make the diffuse measurements unusable.
Data
The
following days were used in the analysis of the comparison: see
http://www.antarctica.ac.uk/met/psa/radiation2003/html/html_code/
http://www.antarctica.ac.uk/met/psa/radiation2004/html/html_code/
year |
month |
day |
"tauhaze" |
2004 |
1 |
28 |
0.03 |
2004 |
2 |
11 |
0.03 |
2004 |
2 |
14 |
0.03 |
2004 |
2 |
15 |
0.02 |
2004 |
2 |
22 |
0.02 |
2004 |
2 |
26 |
0.02 |
2004 |
3 |
10 |
0.015 |
2004 |
12 |
14 |
0.04 |
2004 |
12 |
23 |
0.05 |
The diffuse
radiation, when working on cloud free days, gives an indication of the value of
"tauhaze" used in streamer. This is tuned to
give an agreement across the diffuse data. Once this is found, there is a
residual error with the global radiation. E.g. see Figure 1 for the 10th of
March, 2004. The residual error is due
to error in direct (beam) radiation, and I assume this is due to tilt error in
the global pyranometer.
The tilt can
be estimated by first calculating the Streamer beam radiance. This is the
direct radiation adjusted for the zenith angle of the sun and is equivalent to
the value a virtual pyranometer would read if pointing
directly at the sun at all times. Figure 2 shows the time series of the beam
radiance for the same data in Figure 1.
Figure 1. Global
and diffuse radiation from Streamer and the pyranometers. Although there is good
agreement with the diffuse values, there is a residual gain and phase error in
the global.
Figure 2. Upper panel showing detail of the different in direct radiation between the pyranometers and Streamer for the data in Figure 1. Middle panel the zenith angle of the sun. Lower panel the beam radiation, i.e. the radiation passing though a unit area orthogonal to the beam.
A virtual pyranometer with a tilt error is now used to modify the Streamer beam radiation. The tilt is defined by a pointing vector, P(x,y,z), |P| = 1, which has effectively two free variables: the tilt in the east-west direction and the tilt in the north-south direction. A value of P(x,y,z) is sought to minimise the difference between the virtual Direct radiation and that measured by the real pyranometers.
Figure 3. Upper panel shows the
direct radiation as given by Streamer (black) and by the pyranometers
(blue), with the virtual pyranometer overlaid almost
exactly as red squares. The low panel shows the difference between the measured
and virtual pyranometers.
P(x,y,z)
was found by the Matlab fminsearch
function, minimising the mean magnitude of the difference presented in the
lower panel of Figure 3, and giving
P(x,y,z) =
[0.0070 0.0175 0.9998]
This is equivalent
to an east-west tilt of 0.40o and a north-south tilt of 1.00o.