*******************************************************************
****  SOME VERY IMPORTANT REMARKS CONCERNING THE UV SENSORS *******
****  AND THE INTERPRETATION OF THE OUTPUT OF THESE SENSORS *******
*******************************************************************
12-July-1994        Release 1.1                      Francis MASSEN
                                                     LCD

1. Interpreting the UV_A values
----------------------------------
The values stored in the data-files represent the raw output of the
UV_A sensor. To get the exact total irradiant power (in W/m2), this
number must be modified to respect the non-uniform spectral sensitivity
of the sensor used (a special silicon photodiode, pointing upwards, and
covered by a cosine-diffuser; if the angle of incidence is less than
75 degrees (sun makes angle superior of 25 degrees with respect to the
horizontal), the output of the upwards pointing sensor is independant
of the angle; at 80 degrees, output falls to 86%, at 85 degrees to 68%).
Remember that this output is proportional to the UV_A power falling
on a ***horizontal*** disposed surface! The total power falling on a
surface perpendicular to the sun rays can be calculated by dividing
this output by the cosine of the sun-angle.

As stated in README.TXT, the relative spectral sensistivity response
curve of the UV_A sensor is a slightly asymetric bell curve, which
extends from 320nmm to 400nm (the domain usually defined as the UV_A
range) and peaks at 370nm.

To compute the real UV_A irradiant power falling on a horizontal surface
from the stored values, one has to take into account that the irradiant
power of the solar-light is not quite constant over this wavelength region
(and varies with the seasons,the position of the sun, the atmospheric
conditions ...)

Based on an curve of the solar spectrum  (measured at Munich, Germany)
given by M. Nico Harpes of the Radiation Protection Office of Luxembourg,
and which is valid for a sun-position greater than 60 degrees above horizon,
summer-time, one finds this:

    - total irradiant power in the [320..400nm] range:      35.47 W/m2
    - total UV_A sensor output, if its relative
      spectral response curve is applied to these values:   14.71 W/m2

      Multiplier to use :  35.47/14.71 = 2.411

An example:          the UV_A sensor gives 6 W/m2
                     The total irradiant UV_A power in the [320..400nm]
                     region falling on a *** horizontal surface *** is:
                     irradiant UV_A power = 6 * 2.411 = 14.466 W/m2


If one starts with the simplified hypothesis of an uniform power
distribution of the solar light in the 320 to 400nm region, the multiplier
will be given by the fraction of two areas: a rectangular area
with width (400-320) and height 100, and the area under the bell-shaped
relative spectral response curve, which computes to 3047.
So this multiplier would be 8000/3047 = 2.626 (about 9% more than the
preceding one).

The published UV_A data of the British NRPB (Dr. Driscoll, Solar radiation
measurements at three sites in the UK; NRPB-M452) gives a highest mean
UV_A output for the three NRPB measurement sites (Chilton, 52 latitude North;
Leeds, 54 latitude North; Glasgow, 56 latitude North) in June 1992 of
33 W/m2; this compares to the maximum we measured until now at 24 June 1994
of 8.112 W/m2 ; thus the multipler would be 33/8.112 = 4.068

Preliminary conclusion: the necessary multiplier to convert the stored
                        UV_A data into total irradiant power falling
                        on a horizontal surface lies
                        between 2.411 and 4.07

We did some quick intercomparison with a calibrated instrument of
the Radioprotection Service of Luxembourg (IL1700); the results were not
good due to the varying cloudiness of the sky; they will repeated when
feasible. Nevertheles they suggest that the multiplier to use should lie
between 4 and 6.

I would suggest that until further results are obtained, one should use
the multiplier of 4 as the best guess to convert the logger UV_A output
into total irradiant UV_A power falling on a ***horizontal*** surface
(be aware of the underlying assumptions: sun at angle >60 degrees;
summertime!)

If one sticks to ***relative*** measurements, i.e. uses the UV_A output
just to compare different days or periods at Diekirch to each other, no
correction is needed. Peak values comparison certainly will be the most
reliable to do.

2. The UV_B data:
------------------

As stated in README.TXT, the relative spectral sensitivity response
curve of the UV_B sensor is an almost symetric bell curve, which
extends from 280nmm to 360nm (remember that the domain usually defined
as the UV_B range is 280 to 319nm) and peaks at 313nm.

The solar spectral curve varies over several orders of magnitude in the
280 to 319nm interval (the UV_B region). The hypothesis of an uniform power
distribution can not be applied in any case.

Based on the same curve of the solar spectrum as before, one finds this:

    - total irradiant power in the [280..319nm] range (the
      range covered by the UV_B sensor):                     0.9877 W/m2
    - total UV_B sensor output in the [280..319nm] range
      if its relative spectral response curve is applied to
      the solar spectrum:                                    0.9398 W/m2

      First multiplier to use :  987.7/939.8 = 1.05

As the UV_B region stops at 319nm, a second multiplier representing the
fraction

       (measured irradiant power in [280..319])/(measured irradiant power
       in [280..360])

       = 0.9398/3.722 = 0.253

should be used.

The definitive multiplier to convert the stored UV_B data into irradiant
power in the [280..319nm] range falling on a horizontal surface will be:

   definitive_multiplier = (0.9877/0.9398)*(0.9398/3.722) = 0.2654

An example:  The UV_B sensor output of the LCD station is 4 W/m2
             The true irradiant UV_B power in the [280..320nm] region
             on a ***horizontal*** surface is  4 * 0.2654 = 1.06 W/m2

The same statement concerning relative UV_B measurements as that made
in the UV_A chapter also applies here.
--------------------------------------------------------------------------

3. General conclusions:
---------------------------

The UV_A and UV_B sensors used at the LCD meteorological station are
integrating devices. Making relative comparisons is easy, but getting
meaningful and reliable absolute values from these devices can be a
tricky affair.

What I proposed is a first tentative, yielding two multipliers to get
total irradiant UV_A and UV_B power in W/m2 from the output of the
sensors.

I would be grateful for every suggestions and remarks concerning this
problem.

          francis.massen@menvax.restena.lu

--------------------------------------------------------------- eof -------
Acknowledgments to:       Mr. Nico HARPES
                          Radiation Protection Office, Luxembourg
                          fax: (352) 45 47 94
---------------------------------------------------------------------------

