Collocated time series of narrowband 0.6 μm atmospheric flux transmittance at the surface and bidirectional reflectance at the top of atmosphere are decomposed into distinct frequency bands, so as to investigate the timescale dependences of their variance and correlation. Toward this goal, a multiresolution analysis based on the maximum overlap discrete wavelet transform and the Haar wavelet is applied to 5‐minute resolution measurements from two multifilter rotating shadowband radiometers operated at Cabauw, NL, and Heselbach, DE, and to observations of the geostationary METEOSAT8 SEVIRI satellite imager. Both time series are best correlated when the satellite data are shifted by about 1 pixel (6 km) to the North, which is likely attributable to the parallax effect caused by the height of cloud tops and the oblique satellite viewing angle. While variations in transmittance and reflectance with periods longer than an hour are found to be highly anticorrelated, the correlation breaks down for shorter periods. Below 1 hour, the transmittance time series also exhibits significantly higher variance than the reflectance. The larger extent of the satellite pixel (6 × 3 km2) versus the point nature of the ground measurements is proposed as an explanation. Due to the small contributions of high‐frequency variability to the total variance of the reflectance, aliasing effects caused by the 5‐minute repeat cycle of SEVIRI are expected to be small. Implications of our findings are discussed for the validation of satellite estimates of solar surface irradiance. Averaging of surface measurements over at least 40 minutes is recommended for a comparison.
Read full abstract