The specifications of UV instrumentation are based on the objectives of UV research. These aims include:

 

1. To establish a UV climatology by long-term monitoring, e.g. within a network of UV spectroradiometers

2. To detect trends, especially spectrally resolved tends, in global UV irradiance

3. To provide datasets for specific process studies and for the validation of radiative transfer models and/or satellite derived UV irradiance at the Earth?s surf

4. To understand geographic differences in global spectral UV irradiance

5. To gain knowledge or gather information about actual UV levels

6. To allow the determination of a UV index
 

 

Some of the objectives (e.g. trend detection) require high accuracy instruments with a superior long-term stability because the expected magnitude of UV trends is rather small. In contrast, somewhat less demanding specifications are sufficient for instruments employed for the determination of erythemally weighted UV doses and hence the UV index.

   

Detectability of UV Changes
 

A useful but ambitious goal is the attempt to detect a change in spectral UV irradiance resulting from a 1% change in total ozone column. The primary interest lies in UV increases resulting from reductions in total ozone column. However, possible reductions in UV resulting from future recovery of the ozone layer, or from a build up of tropospheric pollution (e.g. aerosols, ozone) or stratospheric particle loading may be relevant as well.

From past experience, the lowest calibration uncertainty that can be maintained for instruments designed to measure solar UV irradiance appears to be currently limited to a few percent (e.g. ± 5%). Thus, to achieve the above goal, it will be necessary to include measurements at short wavelengths, where small changes in total ozone lead to relatively large changes in UV irradiance.

The spectral changes in UV resulting from a 1% ozone depletion have been calculated for overhead sun and for 70° solar zenith angle (SZA). Relative changes in UV irradiance increase rapidly at shorter wavelengths, but absolute changes decrease at wavelengths shorter than 310 nm. For overhead sun, a radiation change of 5% occurs at approximately 295 nm, when the absolute change in irradiance is approximately 10^-4 Wm^-2nm^-1. At larger SZA, the condition for a 5% change in irradiance occurs at longer wavelengths. However, the corresponding absolute changes are even smaller, and thus more difficult to detect. It should be noted that high-sun observations are not always possible. For example, at high latitudes and in winter or spring, where ozone and UV changes are expected to be largest, the minimum SZA becomes large and can exceed 90°. These considerations show that, given a calibration uncertainty of 5%, the increase in UV resulting from a 1% ozone depletion will be detectable only if the detection threshold is in the order of 10^-6 Wm^-2nm^-1 or lower.

Uncertainties in the measurements of global spectral irradiance resulting from uncertainties in the wavelength alignment escalate at shorter wavelengths. Therefore, in addition to the demands on calibration accuracy and detection threshold, an accurate wavelength alignment is required as well. For example, at 295 nm, a wavelength error of 0.1 nm corresponds to an irradiance error of approximately 9% for 30° solar zenith angle (see below).

In Figure 1, the change in global spectral irradiance due to a 1% and 3% decrease in total ozone column is compared to the uncertainty in spectral measurements arising from a 5% calibration uncertainty, a detection threshold of  10^-6 Wm^-2nm^-1 and a wavelength error of 0.05 nm. For the calculation of spectral irradiance a SZA of 30° and a total ozone column of 300 DU was assumed. The steep increase of the uncertainty at 291 nm results from reaching the detection limit. The figure shows, that the percentage change in UV due to a 1% change in ozone only slightly exceeds the calculated measurement uncertainty at 295 nm. Thus, in order to detect a change of UV radiation caused by a 1% change in ozone, the wavelength alignment accuracy must be significantly better than ±0.05 nm, the detection threshold must be in the order of 10^-6 Wm^-2nm^-1 or lower and the accuracy of the absolute calibration must be at least ± 5%. These values form the basis for the specification of type S-2 instruments. However, it has to be emphasized that the observance of these specifications still does not guarantee that a trend caused by a 1% change in total ozone is detectable[38].

Figure 1: Percentage change in global spectral irradiance caused by a 1% and 3% change of total ozone compared with the uncertainty in spectral measurements arising from a 5% calibration uncertainty, a wavelength error of 0.05 nm, and a detection threshold of 10^-6 Wm²nm^-1. The change in irradiance was calculated for a solar zenith angle of 30° and a total ozone column of 300 DU.

For the validation of radiative transfer models the accuracy of spectral measurements must be comparable to the accuracy needed for trend detection. For the determination of actual UV levels and for the assessment and interpretation of geographical differences in global spectral UV irradiance the accuracy requirements are less demanding, but also require good and well maintained instrumentation.