Development and validation of a novel approach for evaluation of broadband UVA irradiance and total daily UVA exposures from OMI satellite data

Development and validation of a novel approach for evaluation of broadband UVA irradiance and total daily UVA exposures from OMI satellite data

It is important to collect and validate data from satellites in order to obtain global information about the solar UVA (320-400 nm) environment. This research reconstructed and validated the broadband UVA irradiances derived from discrete spectral irradiance data retrieved from the Ozone Monitoring Instrument (OMI) satellite from 1 January to 31 December 2009. OMI data at solar noon was compared to ground based spectral irradiance at Toowoomba (27°36’ S 151°55’ E), Australia at 310, 324 and 380 nm for both cloud free and all sky conditions. There was a strong relationship between the ground based UV spectroradiometer data and satellite based measurements with an R2 of 0.89 or better in each waveband for cloud free days. Models developed for the sub-tropical site data account for these differences and are essential for any correlation between satellite and ground based measurements. Additionally, this research has developed a model to evaluate the solar noon broadband UVA irradiance from the discrete satellite spectral irradiance at 310, 324 and 380 nm, comparing the UVA irradiance at solar noon on cloud free days to those measured over 12 months with a ground based UVA radiometer. An R2 of 0.86 was obtained confirming that for cloud free days the broadband UVA can be evaluated from the OMI satellite spectral measurements.

This research also investigated the influence of cloud on the broadband UVA solar noon irradiance evaluated from the solar noon satellite based OMI spectral UV data that were compared to the ground based radiometer irradiance in a twelve year period, from 1 October 2004 to 31 December 2016. The correlation, calculated with the model, between ground based radiometer data and the evaluated OMI broadband UVA irradiance depend on whether or not the solar disc was obscured by the presence of cloud and on the total sky cloud fraction. For conditions when the sun was not obscured by cloud, the evaluated satellite and the ground-based UVA irradiance correlation was best for cloud cover between 0-2 okta (R2 = 0.78) and worst for high cloud cover of >4 and up to 8 okta (R2 between 0.30 and 0.40). The R2 value reduced with increasing cloud cover and showed significantly weaker correlation when the sun was obscured. The correlation between the evaluated satellite broadband UVA and ground-based measurements over the twelve years for total cloud cover conditions of 4 okta or less confirmed that the broadband UVA satellite evaluation model using the OMI spectral data is valid for approximately 71% of the days at the Southern Hemisphere sub-tropical study site.

This research then developed a method to accurately calculate the total daily broadband UVA exposure from the satellite derived solar noon irradiance on cloud free days. The method utilises cloud free UVA irradiance data, collected daily at high temporal resolution over the period 2005 to 2016, to derive the normalised diurnal UVA exposure and determine a factor relating the solar noon irradiance to the total daily UVA exposure. To demonstrate the versatility of this approach, OMI satellite solar noon UVA irradiance data were employed in calculating the total daily UVA exposures and compared to the respective ground based site measurements. There was a strong correlation between the total daily satellite and ground based broadband UVA exposures (R2 = 0.90). The developed method enables the total daily UVA exposures to be evaluated from satellite solar noon UVA irradiances at sites that do not record short term temporal variations in terrestrial UVA.

daily UVA exposure calculated from the three OMI UV spectral irradiance measures at solar noon. These evaluated satellite total daily UVA exposures were compared to the total daily UVA exposures of a ground based broadband radiometer over the period of October 2004 to December 2014 at the Toowoomba site under all cloud cover conditions including sun obscured and not obscured states. The method was employed to evaluate the influence of cloud on the total daily UVA exposure. When the sun was not obscured by cloud, there was good agreement between satellite and ground based daily UVA exposure with R2 between 0.80 and 0.84 for the cloud conditions 0 to 2, > 2 to 4, > 4 to 6 and > 6 to 8 okta. For sun obscured by cloud, the R2 was 0.71, 0.64 and 0.75 respectively for > 2 to 4, > 4 to 6 and > 6 to 8 okta. The method was validated using total daily UVA exposures from ground measurements taken in 2015 and 2016 giving a mean absolute error of 84.2 kJ/m2 (10%) and 138.8 kJ/m2 (30%) respectively for the cases of sun not obscured cloudy days and sun obscured by cloud cover. Total daily UVA exposures were able to be calculated from the OMI satellite spectral irradiance data for all cloud conditions, including cases when the sun was obscured, demonstrating the potential of the technique to be applied remotely in locations that do not record surface UVA measurements directly.