In this study, we present an aerosol classification technique based on measurements of a double monochromator Brewer spectrophotometer during the period 1998–2017 in Thessaloniki, Greece. A machine learning clustering procedure was applied based on the Mahalanobis distance metric. The classification process utilizes the UV Single Scattering Albedo (SSA) at 340 nm and the Extinction Angstrom Exponent (EAE) at 320–360 nm that are obtained from the spectrophotometer. The analysis is supported by measurements from a CIMEL sunphotometer that were deployed in order to establish the training dataset of Brewer measurements. By applying the Mahalanobis distance algorithm to the Brewer timeseries, we automatically assigned measurements in one of the following clusters: Fine Non Absorbing Mixtures (FNA): 64.7%, Black Carbon Mixtures (BC): 17.4%, Dust Mixtures (DUST): 8.1%, and Mixed: 9.8%. We examined the clustering potential of the algorithm by reclassifying the training dataset and comparing it with the original one and also by using manually classified cases. The typing score of the Mahalanobis algorithm is high for all predominant clusters FNA: 77.0%, BC: 63.9%, and DUST: 80.3% when compared with the training dataset. We obtained high scores as well FNA: 100.0%, BC: 66.7%, and DUST: 83.3% when comparing it with the manually classified dataset. The flags obtained here were applied in the timeseries of the Aerosol Optical Depth (AOD) at 340 nm of the Brewer and the CIMEL in order to compare between the two and also stress the future impact of the proposed clustering technique in climatological studies of the station.

The gap in knowledge regarding the radiative effects of aerosols in the UV region of the solar spectrum is large, mainly due to the lack of systematic measurements of the aerosol single scattering albedo (SSA) and absorption optical depth (AAOD). In the present study, spectral UV measurements performed in Thessaloniki, Greece by a double monochromator Brewer spectrophotometer in the period 1998–2017 are used for the calculation of the aforementioned optical properties. The main uncertainty factors have been described and there is an effort to quantify the overall uncertainties in SSA and AAOD. Analysis of the results suggests that the absorption by aerosols is much stronger in the UV relative to the visible. SSA follows a clear annual pattern ranging from ~0.7 in winter to ~0.85 in summer at wavelengths 320–360 nm, while AAOD peaks in summer and winter. The average AAOD for 2009–2011 is ~50% above the 2003–2006 average, possibly due to increased emissions of absorbing aerosols related to the economic crisis and the metro-railway construction works in the city center.

The gap in knowledge regarding the radiative effects of aerosols in the UV region of the solar spectrum is large, mainly due to the lack of systematic measurements of the aerosol single scattering albedo (SSA) and absorption optical depth (AAOD). In the present study, spectral UV measurements performed in Thessaloniki, Greece by a double monochromator Brewer spectrophotometer in the period 1998 - 2017 are used for the calculation of the aforementioned optical properties. The main uncertainty factors have been described and there is an effort to quantify the overall uncertainties in SSA and AAOD. Analysis of the results suggests that the absorption by aerosols is much stronger in the UV relative to the visible. SSA follows a clear annual pattern ranging from ~0.7 in winter to ~0.85 in summer at wavelengths 320 – 360 nm, while AAOD peaks in summer and winter. The average AAOD for 2009 – 2011 is ~50% above the 2003 – 2006 average, possibly due to increased emissions of absorbing aerosols related to the economical crisis and the metro-railway construction works in the city center. A detailed analysis of the uncertainties in the retrieval of the SSA and the AAOD from the Brewer spectrophotometer has been also performed.

The shortwave downward solar irradiance (SDR) is an important factor that drives climate processes and energy production and can affect all living organisms. Observations of SDR at different locations around the world with different environmental characteristics have been used to investigate its long-term variability and trends at different time scales. Periods of positive trends are referred to as brightening periods and of negative trends as dimming periods. In this study we have used 30 years of pyranometer data in Thessaloniki, Greece, to investigate the variability of SDR under three types of sky conditions (clear-, cloudy- and all-sky). The clear-sky data were identified by applying a cloud screening algorithm. We have found a positive trend of $0.38\%/\mathrm{year}$ for all-sky, ∼0.1%/year for clear-sky, and $0.41\%/\mathrm{year}$ for cloudy conditions. The consistency of these trends, their seasonal variability, and the effect of the solar zenith angle have also been investigated. Under all three sky categories, the SDR trend is stronger in winter, with $0.7$, $0.4$, and $0.76\%/\mathrm{year}$, respectively, for all-, clear-, and cloudy-sky conditions. The next larger seasonal trends are in autumn—$0.42$ and $0.19\%/\mathrm{year}$, for all and cloudy skies, respectively. The rest of the seasonal trends are significant smaller, close to zero, with a negative values in summer, for clear and cloudy skies. The SDR trend is increasing with increasing solar zenith angle, except under cloudy skies, where the trend is highly variable and close to zero. Finally, we discuss shorter-term variations in SDR anomalies by examining the patterns of the cumulative sums of monthly anomalies from the climatological mean, both before and after removing the long-term trend.

Accurate short-term cloud forecasting is essential for solar radiation and solar energy applications, where rapidly evolving cloud fields introduce substantial variability in surface solar irradiance. In this study, we develop a hybrid ConvLSTM-UNet deep learning model for 15-min-ahead forecasting of cloud optical thickness (COT) from Meteosat/SEVIRI observations over Europe. The model combines multiscale spatial feature extraction with recurrent spatiotemporal learning and employs a combined Charbonnier–SSIM loss to improve both numerical accuracy and the preservation of cloud structures. The proposed forecasting model is evaluated against real COT satellite retrievals and benchmarked against persistence and a dense optical flow cloud motion vector (CMV) approach. For 15-min-ahead forecasts, ConvLSTM-UNet achieves lower COT prediction errors and higher perceptual quality than both benchmarks, with MAE and RMSE of 1.25 and 9.52, respectively, and SSIM of 0.95. The impact of COT forecast errors on solar radiation is further assessed through the cloud modification factor (CMF), for which the proposed model achieves a forecast skill of approximately 30 % relative to persistence. Its performance advantage is particularly evident under cloudy and rapidly evolving conditions, and remains evident under recursive forecasts extending to approximately 165 min ahead. These results demonstrate the potential of deep spatiotemporal learning for satellite-based short-term cloud and solar radiation forecasting.

Accurate short-term cloud forecasting is essential for solar radiation and solar energy applications, where rapidly evolving cloud fields introduce substantial variability in surface solar irradiance. In this study, we develop a hybrid ConvLSTM-UNet deep learning model for 15-min-ahead forecasting of cloud optical thickness (COT) from Meteosat/SEVIRI observations over Europe. The model combines multiscale spatial feature extraction with recurrent spatiotemporal learning and employs a combined Charbonnier–SSIM loss to improve both numerical accuracy and the preservation of cloud structures. The proposed forecasting model is evaluated against real COT satellite retrievals and benchmarked against persistence and a dense optical flow cloud motion vector (CMV) approach. For 15-min-ahead forecasts, ConvLSTM-UNet achieves lower COT prediction errors and higher perceptual quality than both benchmarks, with MAE and RMSE of 1.25 and 9.52, respectively, and SSIM of 0.95. The impact of COT forecast errors on solar radiation is further assessed through the cloud modification factor (CMF), for which the proposed model achieves a forecast skill of approximately 30 % relative to persistence. Its performance advantage is particularly evident under cloudy and rapidly evolving conditions, and remains evident under recursive forecasts extending to approximately 165 min ahead. These results demonstrate the potential of deep spatiotemporal learning for satellite-based short-term cloud and solar radiation forecasting.