PARAMETRIC AND EMPIRICAL MODELLING OF SOLAR ULTRAVIOLET RADIATION BASED ON ATMOSPHERIC AND WEATHER PARAMETERS AT BANGI, MALAYSIA

Abstract

Solar ultraviolet radiation (UVR) is essential for the maintenance of life on Earth’s surface. It is important to a wide range of fields, such as medical field, agriculture, and general public health. Therefore, comprehensive and continuous theoretical and experimental research works are required to understand the estimation approach of solar UVR. For the current study, there were three specific objectives: (1) to develop a simple technique for measuring solar UVR within the range of 300–400 nm in Bangi, Malaysia; (2) to characterise the diurnal and seasonal variations of solar UVR on the Earth’s surface at Bangi, Malaysia; (3) to propose key empirical models that can predict solar UVR. A cost-effective measurement system was considered to measure the solar UV intensity within the interval of 300–400 nm from January 2014 to July 2015. All measurements were performed using Avantes Avaspec ULS 2048x64-USB2 spectrometer at the Department of Applied Physics, Universiti Kebangsaan Malaysia, located in Bangi (2° 55’ N, 101° 46’ E, and 50 meters ASL) on the horizontal surface under clear sky condition. Moreover, direct Sun radiation was measured by pointing the sensor directly at the Sun and tracking the shadow. Furthermore, hourly solar UVR intensity was obtained every month, and mean values were calculated. Additionally, parameterised and empirical models were adopted in this study. The first model was based on atmospheric data and used air mass (m) parameter as an independent variable. This model can predict solar UVR on the Earth’s surface and at the top of the atmosphere and then estimate the total atmospheric attenuation. The second model was based on weather data and utilised global solar radiation (G), air temperature (T), and relative humidity (RH) as independent variables. The model can predict solar UVR on the Earth’s surface. The estimated solar UVR from both models were compared with the measured solar UVR data in 2015, which were then validated using the coefficient of determination (R2 ), efficiency test (EFF), root mean square error (RMSE), and mean bias error (MBE). The key output of this study was a measurement system that can collect solar UVR data using a simple, low-cost instrument. The obtained results indicated that solar UVR was at its highest from 11:00 to 13:00 LT (peak hours). The maximum solar UVR was approximately 46.5 Wm-2 (in July). Besides that, seasonal variation showed that solar UVR recorded its highest intensity during the dry season, between April and August. In contrast, the minimum value of solar UVR was 33.71 Wm-2 , which was recorded in November. Moreover, the rainy season from October to March recorded the lowest solar UVR lower intensity. On the other hand, significant correlations of m, G, T, and RH with UVR were established in this study. The proposed equations in this study successfully predicted the values of UVR, as R2 exceeded 0.8 for most of the months. Additionally, both models provided satisfactory validation results; the recorded R 2 and EFF were close to unity. The negative values of MBE for most models showed that the predicted data were below the measured data, except for temperature model and relative humidity model. Besides that, this study recorded less than 3% for MBE%, less than 6 for RMSE values, and below 19% for RMSE%, showing low and acceptable dispersion indicators. However, a comparison of R2 showed that the parameterised model produced better outcomes than the empirical model. In conclusion, this study achieved all objectives with the development of a cost effective observation system and data measurement as well as model construction