See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/50265618

Know Your Standard: Clarifying the CIE Erythema Action Spectrum

Article  in  Photochemistry and Photobiology · March 2011

DOI: 10.1111/j.1751-1097.2010.00871.x · Source: PubMed

CITATIONS READS

103 3,617

4 authors, including:

Harry Slaper Peter Koepke

National Institute for Public Health and the Environment (RIVM) Ludwig-Maximilians-University of Munich
77 PUBLICATIONS   3,584 CITATIONS    125 PUBLICATIONS   7,073 CITATIONS   

SEE PROFILE SEE PROFILE

Alois Schmalwieser
University of Veterinary Medicine, Vienna
104 PUBLICATIONS   1,339 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Forecast of the erythemally and other effective solar UV radiations View project

Radon, thoron and other sources of natural radiation in the Netherlands View project

All content following this page was uploaded by Alois Schmalwieser on 02 February 2019.

The user has requested enhancement of the downloaded file.

Photochemistry and Photobiology, 2011, 87: 483–486
Research Note
Know Your Standard: Clarifying the CIE Erythema Action Spectrum
Ann R. Webb*1, Harry Slaper2, Peter Koepke3 and Alois W. Schmalwieser4
1
School of Earth Atmospheric and Environmental Sciences, University of Manchester, Manchester, UK
2
Laboratory for Radiation Research, The National Institute for Public Health and the Environment (RIVM),
Bilthoven, The Netherlands
3
Meteorological Institute, Ludwig-Maximilians-University, Munich, Germany
4
Unit of Molecular Physiology and Biophysics, University of Veterinary Medicine, Vienna, Austria
Received 24 August 2010, accepted 23 November 2010, DOI: 10.1111/j.1751-1097.2010.00871.x
ABSTRACT
The standard erythema action spectrum provides an internation-
ally accepted representation of the erythema-inducing effective-
ness of wavelengths in the UV part of the spectrum. The action
spectrum forms the basis of the UV index used for public health
information, defines the standard erythema dose unit and the
minimum erythema dose and is the default response spectrum
aspired to by a range of UV radiometer manufacturers. However,
there are several versions of this erythema action spectrum in use,
and only one of them has been endorsed as a standard. While the
differences in erythemally weighted radiation incurred by choice of
action spectrum will be no more than a few percent, this uncertainty
is unnecessary. Here we detail the differences in the different
versions of erythema action spectra, illustrate the resulting effects
in quantifying UV doses and encourage readers to use only the
standard version of the action spectrum in the future.
INTRODUCTION
The erythema action spectrum is widely used in assessing the
negative effects of UV radiation sources (most frequently the
sun) on human skin. Strictly, the action spectrum is for
erythema (sunburn), but it is also used as a proxy in
quantifying UV for other UV effects, e.g. skin cancer, vitamin
D synthesis and even effects on plants and sometimes for
weathering. The action spectra for these effects are similar but
often not precisely known. Thus, many UV radiometers are in
use that measure the erythemal UV directly with a single
channel that has a response mimicking the erythema action
spectrum (1). This is in part because it takes more detailed
spectral measurements, which need more complex instruments,
to precisely define alternative biologically weighted radiation.
The erythema action spectrum is also the basis of the UV index
(2,3), which is used globally in weather forecasting and for
public health information. The UV index is simply erythemally
effective UV scaled to give a dimensionless number between 1
and 12 (or more in extreme cases) that indicates the instan-
*Corresponding author email: ann.webb@manchester.ac.uk (Ann R. Webb)
 2011 The Authors
Photochemistry and Photobiology  2011 The American Society of Photobiology 0031-8655/11
483
taneous sunburning potential of the solar radiation. The total
erythemal radiation received can be expressed in terms of
standard erythema dose unit (SED) where 1 SED is defined as
100 J m)2 erythemally effective radiation (4). The minimum
erythema dose (MED) is also based upon the erythema action
spectrum, being the minimum dose of erythemal radiation that
produces a faint but perceptible reddening of the skin 24 h
after irradiation (5). However, the MED is not a standardized
term but a personal measure of susceptibility to sunburn.
Thus, both common instruments and derived units are based
on the erythema action spectrum.
The erythema action spectrum has come to refer, through
common use, to the action spectrum proposed by McKinlay
and Diffey (6,7). The action spectrum is a mathematical fit to
experimental data and is represented by three straight lines
on a semilog plot. It was the work of a Technical Committee
of the International Commission on Illumination (CIE),
intended to replace existing ‘‘Standard erythemal curves’’
from CIE, 1935 (8) and DIN (9), which did not account for
the UVA portion of the spectrum. In addition to information
in the UVA, the ‘‘new’’ erythema action spectrum was based
on statistical considerations of (what was then) relatively
recent experimental data, and it was deemed to represent a
good ‘‘average erythema curve’’—recognizing that there is
inevitable variation in human skin and its response to
radiation.
This CIE action spectrum gained broad acceptance and
widespread use, and in 1998 was published as a CIE standard
(10) and thereafter adopted by ISO (11)—we will refer to this
version as CIE 1998 throughout. However, it is not widely
recognized that there are subtle differences between the 1987
and 1998 versions of the CIE erythema action spectrum.
Here we identify those differences, discuss their implications
and reiterate the standard form of the erythema action
spectrum.
MATERIALS AND METHODS
The 1987 versions. In the original publications (6,7) the authors showed
a table for the erythemal weighting function with values with two
significant digits (see Table 1, column 3) and the following set of
equations (Eq. [1]).
484 Ann R. Webb et al.

Table 1. The erythema action spectrum as represented by tables and calculations based on McKinlay and Diffey 1987, Diffey 1994 ⁄ DIN5031–10
(Eq. 1) and CIE 1998 (Eq. 2). The percentage differences between different versions of the action spectrum are also represented.

ery(k) Percentage Percentage Percentage Percentage

Eq. (1) ery(k) CIE 87 (Eq. 1) CIE87 (Eq. 1) Eq. (2) CIE98 (Eq. 2)
ery(k) table ery(k) table modified Eq. (2) and CIE98 and table and table and table
k (nm) CIE 1998 McK&D1987 CIE87 CIE98 (Eq. 2) McK&D87 CIE98 McK&D87

298 1.0000000 1.00000 1.0000000 1.0000000 0.0 0.0 0.0 0.0

300 0.64863 0.65 0.64863 0.64863 0.0 )0.2 0.0 )0.2
310 0.074473 0.074 0.074473 0.074473 0.0 0.6 0.0 0.6
320 0.008551 0.0086 0.008551 0.008551 0.0 )0.6 0.0 )0.6
327 0.0018793 0.0018793 0.0018793 0.0 )22.1* 0.0 )22.1*
328 0.0015136 0.0015136† 0.0015136 0.0 )24.8* 0.0 )24.8*
329 0.0014622 0.0014125 0.0014622 )3.4 )12.9* 0.0 )12.9*
330 0.0014125 0.00140 0.0013646 0.0014125 )3.4 )2.5 0.0 0.9
331 0.0013646 0.0013183 0.0013646 )3.4 0.0
339 0.0010351 0.0010000 0.0010351 )3.4 0.0
340 0.0010000 0.00097 0.0009661 0.0010000 )3.4 )0.4 0.0 3.1
350 0.0007079 0.00068 0.0006839 0.0007079 )3.4 0.6 0.0 4.1
360 0.0005012 0.00048 0.0004842 0.0005012 )3.4 0.9 0.0 4.4
370 0.0003548 0.00034 0.0003428 0.0003548 )3.4 0.8 0.0 4.4
380 0.0002512 0.00024 0.0002427 0.0002512 )3.4 1.1 0.0 4.7
389 0.0001841 0.0001778 0.0001841 )3.4 0.0
390 0.0001778 0.00017 0.0001718 0.0001778 )3.4 1.1 0.0 4.6
399 0.0001303 0.0001259 0.0001303 )3.4 0.0
400 0.0001259 0.00012 0.0001216 0.0001259 )3.4 1.3 0.0 4.9

*A log-linear interpolation has been used to interpolate between tabulated values, provided every 10 nm. †Value at 328 nm undefined. Could be
0.0014622 or 0.0015136.

eryðkÞ ¼ 1:0 250 nm  k  298 nm
eryðkÞ ¼ 100:094 ð298kÞ 298 nm  k  328 nm ð1Þ
eryðkÞ ¼ 100:015 ð139kÞ 328 nm  k  400 nm
Due to the £ symbols, the erythema weighting value at 328 nm is
not really defined, since the two equations that refer to this wavelength
give different results.
In a later publication from Diffey (12) and in the updated German
standard, DIN5031 (13) the equations were slightly altered to avoid
the overlapping wavelength intervals. In those publications the first
‘‘equal or smaller sign’’ in each line of Eq. (1) was replaced by a
‘‘smaller’’ sign. This removes the undefined value at 328 nm, but it
does mean that a small downward jump of 3.4% in sensitivity occurs
just above 328 nm, since lines 2 and 3 in the equations still do not
match near 328 nm. For most applications the weighting functions
based on the original and modified description give identical results
(except at 328 nm). We will refer to weighted erythemal doses using the
altered Eq. (1) as CIE87. In addition, in DIN5031 (13) a further
tabulation of the values is provided. These values are based on the
modified Eq. (1), except for the value at 328 nm which is calculated
using the second line in Eq. (1) instead of the third line.
The 1998 standard. The 1998 CIE standard (11) uses a modified
form of line 3 in Eq. (1), such that the constant 139 in the exponent
becomes 140, and affects all wavelengths greater than 328 nm (Eq. 2).
At wavelengths of 328 nm and less there is no change.
eryðkÞ ¼ 1:0 250 nm<k  298 nm
eryðkÞ ¼ 100:094 ð298kÞ 298 nm<k  328 nm ð2Þ
eryðkÞ ¼ 100:015 ð140kÞ 328 nm<k  400 nm
This has the effect of bringing the tabulated values of CIE 1998 (11)
and the associated formulae (Eq. 2) into agreement. Values calculated
with Eq. (2) are shown in Table 1, column 5.
It should be noted that using a log-linear interpolation of the tabled
values at 10 nm wavelength steps as given in the initial publications by
McKinlay and Diffey (6,7) can lead to significant deviations near
328 nm (see Table 1, column 7).
To illustrate the consequences of different versions of the action
spectrum in determining solar UV index readings, spectrally resolved
ambient solar UV-measurements are used which were obtained from
the Dilor XY UV-monitoring system at Rijks Instituut voor Volks-
gezondheid en Milieu, National Institute for Public Health and the
Environment (RIVM), Bilthoven (14). The spectra are analyzed using
the SHICrivm data analysis tool (15), which is used to calculate the
erythemal weighting according to Eqs. (1) and (2) for over 250 spectral
readings obtained on 4 days in 2009 with varying total ozone columns.
Ozone data were obtained from Brewer MkIII measurements from
Koninklijk Nederlands Meteorologisch Instituut, Royal Netherlands
Meteorological Institute (KNMI) in de Bilt (<3 km from the RIVM
site).
RESULTS
The effects of the stepwise changes in representation of the
UVA end of the erythema action spectrum are shown in
Table 1, which also indicates the percentage differences (at a
given wavelength) between different combinations of pub-
lished tables and equations.
The only unequivocal agreement is between the table and
equations (Eq. [2] in this publication) in the 1998 CIE
standard. However, at wavelengths >328 nm this differs
systematically (is 3.4% greater at all wavelengths) from earlier
versions of the action spectrum.
Note that at wavelengths <328 nm where the radiation
induces the greatest erythemal response, all published versions
of this action spectrum agreed. The uncertainty in erythemal
effectiveness of a radiation source, associated with the choice of
erythema action spectrum, will depend on the relative amounts
of radiation greater and less than 328 nm in the source spectrum.
For a source with no output <328 nm, the uncertainty is 3.4%;
for a source with no output >328 nm the uncertainty is
zero. For UV sources with spectral components on both sides
of the 328 nm divide, the uncertainty will be somewhere between
these two extremes.
Such a situation applies with the sun, and since the
erythema action spectrum is widely used in public health
 Photochemistry and Photobiology, 2011, 87 485

1 action spectra are only 0.4–0.5%. For SZA <65 the uncer-
tainties remain less than 1%. For larger SZA the differences
0.995 increase to around 2% at 85.
Ratio - CIE87/CIE98

0.99
0.985
Day 185 300 DU
0.98 Day 162 396 DU
Day 288 246 DU
Day 117 390 DU
0.975
0 10 20 30 40 50
SZA
Figure 1. Ratios of erythemally weighted solar spectra measured in
Bilthoven on 4 days in 2009 with different column ozone amounts. The
measured spectra were weighted with CIE 1987 (modified Eq. [1]) and
CIE 1998 (Eq. [2]). The ratio, taken as CIE 1987 ⁄ CIE 1998, was
plotted against the SZA at the time of measurement.
applications we show (Fig. 1) the differences in erythemally
weighted UV, using Eqs. (1) and (2), for a range of combi-
nations of solar zenith angle (SZA) and column ozone (the two
main determinants of the UV solar spectrum). Measured
spectra for 4 days with different column ozone amounts were
weighted with each of the two alternative action spectra and
the ratios are plotted in Fig. 1 as a function of SZA. If, as a
first approximation the path length through the ozone layer is
assumed to be proportional to the thickness of the ozone layer
and inversely proportional to the cosine of the SZA, and we
assume Lambert–Beers absorption law we can align the results
for different ozone columns by an effective shift in SZA. We
find:
SZA0 ¼ arccosð½D0=D1  cos½SZA1Þ
where D0 is the ozone column for the normalization, taken
here as 300 DU, D1 is the actual total ozone column in DU,
SZA1 is the true SZA at the time of measurement and SZA0 is
the aligned value.
If this transformation is performed on the data from Fig. 1,
we get the results normalized for ozone in Fig. 2. Now the
discrepancies between erythemally effective radiation because
of the different action spectra can be discussed in terms of SZA
alone. At SZA <40 the deviations resulting from the two
1
0.995
Ratio - CIE87/CIE98
DISCUSSION
The so-called McKinlay–Diffey (6,7) and CIE (7,11) erythema
action spectra and other versions (12,13) are widely used and,
one suspects, indiscriminately if unwittingly assumed to be
identical. While there are some small differences between the
ISO ⁄ CIE standard (10,11) and the original publication from
60 70 80 90 100 which it evolved (7), the differences are well within the
uncertainty of the biological data on which the action
spectrum was based. The original authors state that there is
no reason to favor one form of the action spectrum over the
other, either could be supported by the original data and there
is such variation in the human response that the action
spectrum can never be exact (B. L. Diffey and A. F. McKinlay,
personal communication).
In public health applications, for which we take the UV
index as the appropriate unit, differences in choice of erythema
action spectrum will result in uncertainties of no more than
about 2%. Since the dissemination of UV index to the public is
by integer values, this will have negligible effect, and is anyway
well within other uncertainties (e.g. measured or modeled UV
values) used in deriving the UV index.
Of greater concern is the impact that the choice of action
spectrum has on the calibration and comparison of instruments
which measure erythemally weighted UV. Two perfect instru-
ments that use Eqs. (1) and (2), respectively as a reference
action spectrum would differ by up to 2% when measuring the
sun. As shown in Table 1, measuring a source emitting only
wavelengths >328 nm would lead to the extreme case of a
ð3Þ difference of up to 3.4%. In reality the perfect instrument does
not exist and this 2–3% becomes an additional source of
uncertainty in instrument intercomparisons, or in validating
satellite-derived UV. Clarifying and promulgating a single
version of the erythema action spectrum would remove this
unwanted uncertainty.
Even though the uncertainties arising from different ver-
sions of the erythema action spectrum are relatively small, this
is not desirable in a quantity which is considered ‘‘standard-
ized.’’ In such circumstances the internationally endorsed
standard erythema action spectrum, as represented in both
formula and tabulated style in CIE 1998 (10) and ISO ⁄ CIE
1999 (11)—Eq. (2) in this article—should be used in all
applications which refer to the standardized erythema action
spectrum.

Acknowledgements—We thank Alastair McKinlay and Brian Diffey

0.99 for their comments.

0.985
Day 185 300 DU
Day 162 396 DU REFERENCES
Day 288 246 DU
0.98 1. World Meteorological Organization (2006) Instruments to Mea-
Day 117 390 DU
Poly. (Day 185 300 DU)
sure Solar Ultraviolet Radiation, Part 2: Broadband Instruments
0.975
Measuring Erythemally Weighted Solar Irradiance. WMO TD
No. 1289, WMO ⁄ GAW No. 164. WMO ⁄ GAW, Geneva,
0 10 20 30 40 50 60 70 80 90 100
Switzerland.
SZA (corrected for 300 DU)
2. World Meteorological Organization (1997) Report of the WMO-
Figure 2. The data from Fig. 1 transformed according to Eq. (3) to WHO Meeting of Experts on Standardization of UV Indices
account for column ozone amounts. and their Dissemination to the Public. WMO ⁄ TD-No.921,
486 Ann R. Webb et al.
WMO ⁄ GAW Report No. 127. WMO ⁄ GAW, Geneva,
Switzerland.
3. Commission Internationale de l’Eclairage (2003) International
Standard Global Solar UV Index. CIE Standard S 013:2003. CIE
Central Bureau, Vienna, Austria.
4. Commission Internationale de l’Eclairage (1997) Standard
Erythema Dose, a Review. Technical Report CIE 125-1997. CIE
Central Bureau, Vienna, Austria.
5. Epstein, J. H. (1962) Polymorphous light eruptions:
Wavelength dependency and energy studies. Arch. Derm. 85, 122–
128.
6. McKinlay, A. F. and B. L. Diffey (1987a) A reference action
spectrum for ultra-violet induced erythemal in human skin. In
Human Exposure to Ultraviolet Radiation: Risks and Regulations
(Edited by W. F. Passchier and B. F. Bosnjakovic), pp. 83–87.
Elsevier, Amsterdam, The Netherlands.
7. McKinlay, A. F. and B. L. Diffey (1987b) A reference action
spectrum for ultraviolet induced erythemal in human skin. CIE J.
6, 17–22.
8. Commission Internationale de l’Eclairage (1937) Compte Rendu 9
(Tagung Berlin ⁄ Karlsruhe Juli 1935), pp. 596–625. Cambridge
University Press, Cambridge, UK.
9. Deutsches Institut für Normung (1979) DIN5031-10 Vornorm:
Strahlenphysik im Optischen Bereich und Lichttechnik, Größen,
Formel- und Kurzzeichen für Photobiologisch Wirksame Strahlung
(Optical Radiation Physics and Illuminating Engineering; Quanti-
ties and Symbols of Photobiologically Effective Radiation).
Deutsches Institut für Normen, Beuth Verlag Berlin ⁄ Köln,
Germany.
View publication stats
10. Commission Internationale de l’Eclairage (1998) Erythema
Reference Action Spectrum and Standard Erythema Dose. CIE
S007E-1998. CIE Central Bureau, Vienna, Austria.
11. International Organization for Standardization ⁄ Commission
Internationale de l’Eclairage (1999) Joint ISO ⁄ CIE Standard:
Erythema Reference Action Spectrum and Standard Erythema
Dose. ISO 17166 ⁄ CIE S007 ⁄ E-1999. CIE Central Bureau, Vienna,
Austria.
12. Diffey, B. L. (1994) Observed and predicted minimal
erythema doses: A comparative study. Photochem. Photobiol. 60,
380–382.
13. Deutsches Institut für Normung (2000) DIN5031-10 Strahlen-
physik im Optischen Bereich und Lichttechnik, Teil 10: Photo-
biologisch Wirksame Strahlung, Größen, Kurzzeichen und
Wirkungsspektren (Optical Radiation Physics and Illuminating
Engineering—Part 10: Photobiologically Effective Radiation,
Quantities, Symbols and Actions). Deutsches Institut für Normen,
Beuth Verlag Berlin ⁄ Köln, Germany.
14. Den Outer, P. N., H. Slaper and R. Tax (2005) UV radiation in the
Netherlands: Assessing long-term variability and trends in relation
to ozone and clouds. J. Geophys. Res. 110, D02203–D02211.
15. Slaper, H. and T. Koskela (1997) Methodology of intercomparing
spectral sky measurements, correcting for wavelength shifts, slit
function differences and defining a spectral reference. In The
Nordic Intercomparison of Ultraviolet and Total Ozone Instruments
at Izana, October 1996 (Edited by B. Kjeldstad, B. Johnson and
T. Koskela), pp. 89–108. Finnish Meteorological Institute,
Helsinki. (SHICrivm: Available at: http://www.rivm.nl/shic.
Accessed on Nov. 2010.)