Mapping bio-UV products from space

By: Michael Taylor

Solar radiation arriving at the Earth’s surface in the UV part of the spectrum modulates photosynthetically-sensitive life on the land and in the oceans. UV radiation also drives important chemical reaction pathways in the atmosphere that impact air quality. It can cause DNA-damage in the epithelial cells of our skin and is a key factor for tuning the rate of Vitamin D production in our metabolism.

Solar UV radiation may be measured in radiometric units or spectrally-weighted to account for biologically-effective UV radiation doses. The Commission Internationale de l’Éclairage (CIE) defines the reference action spectrum for the ability of UV radiation as a function of wavelength to produce just perceptible erythema (colour change from the Greek word “ερυθρός” for red) in human skin. The standard erythemal dose (Jm-2) is equivalent to an erythemal radiant exposure of 100 Jm-2 (ISO 17166:1999). According to the Bunsen-Roscoe law of reciprocity (Bunsen & Roscoe, 1859), a given biological effect due to UV radiant exposure is directly proportional to the total energy dose given by the product of irradiance (Wm-2) and exposure time (s).

Figure 1: TEMIS erythemal UV dose products (kJ m-2) from KNMI/ESA (Van Geffen et al, 2017): daily “Clear sky” UV from SCIAMACHY/GOME-2, daily “cloud-modified” UV from SCIAMACHY/GOME-2 revealing the impact of a weather system over Sicily, and the global climatological “clear sky” June mean from GOME showing the impact of desert dust as revealed (lower right) by the global climatology of aerosol mixtures (Taylor et al, 2015).

Satellites like GOME, GOME-2 and SCIAMACHY have operational processing algorithms that retrieve erythemal UV dose (kJ m-2) once daily from top of the atmosphere irradiance measurements which are strongly affected by both cloud and atmospheric aerosol (Fig. 1).

Recent studies performed in the context of solar energy (Kosmopoulos et al., 2017; 2018), have revealed that atmospheric aerosol, and desert dust in particular, strongly attenuates solar radiation and the UV component arriving at the ground. In the context of increasing our global capacity for renewable energy with solar power as a major component, this is important. Since atmospheric aerosols reduce solar radiation by absorbing and scattering light and reduce the strength of the direct beam from which solar power generation is most efficient, they also cause forecast uncertainty. As a result, electricity supplied to national grids from solar power must balance the demand by coping with these unexpected fluctuations.

Figure 2: (a) Window functions used by KNMI/ESA to derive Bio-UV dose products from UV spectral irradiances and (b) the back-propagation neural network used to perform ground-based validation. See Zempila et al (2017) for details.

Nevertheless, it is straightforward to obtain biological ultraviolet (“Bio-UV”) products from the UV spectra retrieved at the ground. By applying weighting functions to the ultraviolet part of the irradiance spectra over the range 285-400 nm, important Bio-UV products like Vitamin D dose, DNA-damage dose and photosynthetically active radiation can be calculated. Satellite Bio-UV products from TEMIS (KNMI/ESA) have been successfully validated with high temporal resolution (1-minute) ground-based measurements by Zempila et al. (2017). Fig. 2 shows how the weighting functions vary with wavelength together with the neural network model developed to convert combinations of UV irradiances (I) and solar zenith angle (sza) to Bio-UV products for the ground-based validation.

Figure 3: Zoom sequence showing how the surface solar radiance spectra (280-2500 nm) retrieved under cloudy conditions from space with fast neural network radiative transfer solvers (Taylor et al., 2016) can be used to extract erythemal UV spectral irradiances for calculation of Bio-UV products as per Zempila et al., (2017).

While polar orbiting satellites like GOME, GOME-2, SCIAMACHY and OMI allow global maps of Bio-UV products to be generated, geostationary satellites like Meteosat Second Generation (MSG) provide high spatial resolution images of the Earth disc (3 km x 3 km) every few minutes and allow us to dramatically increase the frequency of the data. In support of this, an operational algorithm capable of retrieving the UV part of the solar spectrum at the surface was recently developed (Taylor et al., 2016). This was achieved with a synergistic model that uses both machine learning with neural networks and a look-up table of radiative transfer simulations to help unravel the complexity of the atmosphere. The model includes the effects of clouds, aerosols, ozone, elevation and surface albedo (Taylor et al; 2016; Kosmopoulos et al., 2017) and provides the surface global horizontal irradiance (GHI), direct normal irradiance (DNI) and diffuse horizontal irradiance (DHI) spectrum over the broad wavelength range 285-2700 nm. Application of the weighting functions of Fig. 2 to the UV part of the solar radiation spectrum can then provide global maps of Bio-UV products at high frequency. Fig. 3 illustrates how various UV products can be obtained from surface solar radiation spectra retrieved from space.

One of the most exci­ti­ng appl­icati­ons of being able to map UV spectral information ­from space is the potential for creating mob­ile appli­cat­ions that pull data from surface UV spectra data cloud computing resources and combi­ne them wi­th users’ GPS i­nformati­on to produce real-ti­me UV alerts to the general publi­c with unprecedented precision. By improving our capacity to map UV impact on the quality of life in the global ecosystem from space at high frequency, we will be better placed to monitor progress towards achievement of the UN’s sustainable development goals as we proceed to a more climate resilient society.


Bunsen, R., Roscoe, H. E., 1859: Photochemische untersuchungen. Annalen der Physik  184(10), 193-273, DOI: 10.1002/andp.18591841002

Kosmopoulos, P., S. Kazadzis, H. El-Askary, M. Taylor, A. Gkikas, E. Proestakis, C. Kontoes, M. M. El-Khayat, 2018: Earth-Observation-Based Estimation and Forecasting of Particulate Matter Impact on Solar Energy in Egypt. Remote Sens. 10(12), 1870, DOI:

Kosmopoulos, P. G., S. Kazadzis, M. Taylor, E. Athanasopoulou, O. Speyer, P. I. Raptis, E. Marinou, E. Proestakis, S. Solomos, E. Gerasopoulos, V. Amiridis, 2017: Dust impact on surface solar irradiance assessed with model simulations, satellite observations and ground-based measurements. Atmos. Meas. Tech., 10(7), 2435-2453, DOI: 10.5194/amt-10-2435-2017

Taylor, M., P. G. Kosmopoulos, S. Kazadzis, I. Keramitsoglou, C. T. Kiranoudis, 2016: Neural network radiative transfer solvers for the generation of high resolution solar irradiance spectra parameterized by cloud and aerosol parameters. J. Quant. Spectrosc. Radiat. Transfer, 168, 176–192, DOI: 10.1016/j.jqsrt.2015.08.018

Taylor, M., S. Kazadzis, V. Amiridis, R. A. Kahn, 2015: Global aerosol mixtures and their multiyear and seasonal characteristics. Atmos. Environ., 116, 112–129, DOI: 10.1016/j.atmosenv.2015.06.029

Van Geffen, J., Van Weele, M., Allaart, M. and Van der A, R., 2017: TEMIS UV index and UV dose operational data products, version 2, KNMI Dataset, DOI:

Zempila, M. M., J. H. van Geffen, M. Taylor, I. Fountoulakis, M. E. Koukouli, M. van Weele, R. J. van der A, A. Bais, C. Meleti, D. Balis, 2017: TEMIS UV product validation using NILU-UV ground-based measurements in Thessaloniki, Greece. Atmos. Chem. Phys., 17(11), 7157–7174, DOI: 10.5194/acp-17-7157-2017


I am very grateful to colleagues from the Tropospheric Emission Monitoring Internet Service (TEMIS) at KNMI and ESA for kindly making available plots of UV radiation monitoring products generated from the v2 processing algorithm, and to colleagues from the Greek national network for the measurement of ultraviolet solar radiation ( for permission to present results from Zempila et al., (2017) using their ground-based NILU-UV multi-filter radiometer measurement data and associated UV dose data obtained from a Brewer MKIII spectrophotometer. I would also like to acknowledge colleagues from with whom I collaborated with to develop the solar radiation neural network modeling aspects presented.

This entry was posted in Climate, earth observation, Remote sensing, Solar radiation. Bookmark the permalink.