Surface UV radiation monitoring based on GOME and SCIAMACHY

Jos van Geffen, Ronald van der A, Michiel van Weele, Marc Allaart and Henk Eskes
in: Proceedings of the ENVISAT & ERS Symposium, 6-10 September 2004,
Salzburg, Austria, ESA publication SP-572, 2005 (CD-ROM).

6. Concluding remarks

As a support to UV radiation monitoring, the TEMIS project provides as data products the UV index and UV dose based on ozone observations by the GOME and SCIAMACHY instruments. The paper outlined the method used to derive these quantities and gave an overview of the available and planned data products.

Section 4.1 presented a preliminary validation of the erythemal UV dose. A more detailed validation is necessary, initially based on cloud-free dates. The EDUCE database is of great importance for this validation, as it provides UV spectra measured on a variety of European locations. For this validation, individual measurements at specific times of the day will be compared, as well as daily total and monthly average values.

A more detailed validation will have to take into account characteristics of the groundbased instruments, such as a dependency of the measurements on the solar zenith angle (which may cause a difference of a few percent), the aerosol amount and the ozone column. Furthermore, the groundbased instruments usually do not measure at wavelengths all the way up to the 400 nm relevant for the erythemal UV index and UV dose. Instruments such as the double monochromator Brewer spectrometers at e.g. De Bilt, Lampedusa and Thessaloniki measure up to 365 nm, and a correction must be made for the range between 365 and 400 nm. This causes only a small difference, since the impact of the missing UV-A wavelengths on the erythemal UV index and UV dose is small (cf. the action spectrum in Fig. 1). [12,13]

A validation of the world-wide UV dose data can only be performed on the basis of monthly averages, as the ISCCP cloud data base used for determining the world-wide UV dose has only monthly-averaged cloud cover fields. Such a validation will therefore be of limited value only.

As mentioned in Section 3.2, the parametrisation used for the UV index implicitly contains the average aerosol load in De Bilt and Paramaribo, hence the TEMIS algorithm currently only contains a "zero-order" aerosol correction. For situations in which there are clearly more aerosols or less aerosols than the average in De Bilt and Paramaribo, a proper aerosol correction is needed, as the preliminary validation of the erythemal UV dose clearly showed. Some work on this has been done by Bodesa and Van Weele [11] and their method will be implemented in the TEMIS algorithm at a later stage.

The TEMIS algorithm for the daily UV dose for Europe currently uses the cloud cover fraction derived from the 1-hourly METEOSAT observations. An improvement would be to use data from the Meteosat Next Generation (MSG) satellites, which is available every 15 minutes. It may then also be possible to further improve the cloud cover correction by using information on the cloud height, available in the MSG data. Furthermore, it would be useful if all data of MSG and other geostationary satellites can be used to determine a UV dose which covers the whole world on a daily base, rather than having to resort to the monthly-averaged data of the ISCCP database.

The TEMIS data service will be continued in the future, extending beyond the end of the project (December 2004), providing an archive of UV data based on GOME and SCIAMACHY measurements, as well as a forecast of the erythemal UV index for today and a few days ahead for the whole world. A reprocessing of these data will take place when either the input global ozone data fields are updated or when the TEMIS algorithm is improved.

Data from GOME spans a period of nearly 8 years, which is insufficiently long to do trend studies. A combination of GOME and SCIAMACHY based UV data is therefore necessary. As both the ozone and the UV data are computed with the same algorithm for both instruments, a combination of data records may provide enough data for trend analysis. But since the two instruments are intrinsically different, such a combination must be done with care, so as not to introduces artificial features.

The UV index and UV dose presented here cover the UV-A and UV-B wavelength ranges. For some biological research fields it may be useful to split this and have index and dose values for UV-A and UV-B separately. Though such a devision is certainly possible when deriving the parametrisation of the UV index as function of total ozone column and solar zenith angle, the resulting parametrisations will have larger uncertainties, as the parametrisation is based on fewer data points (wavelengths) than when using the full wavelength range.

The UV data currently provided is based on two action spectra, namely those for erythema (reddening of the skin due to sunburn) and generalised DNA-damage, as discussed in Section 2. Other action spectra can be used as well for a parametrisation of the UV index. This could be useful, as each action spectrum has a different wavelength dependency, thus giving more information as to how changes in the ozone concentration in the stratosphere affect the UV irradiance at the Earth's surface, and the possible biological effects of these changes.

For example the mammalian non-melanoma skin cancer action spectrum [14], which lies between the erythemal and DNA-damage action spectrum (see Fig. 1), but shows a sine-like wavelength dependency between 340 and 400 nm. Another example is the action spectrum for melanoma induction in platyfish-swordtail hybrids [15], which has a much more significant UV-A component, indicating that ozone depletion would not have as great an influence on melanoma in these species of fish as would the responses described by other action spectra. For references to more action spectra, see Refs. [1,2].

Note that the action spectra in themselves only give an indication of the relative wavelength dependency of biological effects: the actual biological response is determined by the actual dose amount, i.e the UV irradiance weighted with the action spectrum and integrated over the wavelength range and the exposure time, keeping in mind that the dose-response relation may not be linear.

<=== contents of this paper
===> references

<=== Publications page
<=== Post-doc. research at KNMI page

Jos van Geffen -- Home  |  Site Map  |  Contact Me

created: 3 September 2004
last modified: 1 September 2005