Abstract
Natural illumination undergoes changes in hue and saturation associated with sun elevation, yet people do not seem to perceive these changes directly. Previously, we measured speed detection thresholds of smooth changes in daylight metamers and showed that discrimination of chromaticity changes depends on the Correlated Color Temperature (CCT) of the base (adapting) illumination: cool-changes become less noticeable for progressively warmer base lights and vice-versa (Pastilha et al., 2020). Here we analyze temporal variations of real daylight. We aim to determine the velocity range of daylight chromaticity changes and compare it to the previously found discrimination thresholds (Pastilha et al., 2020). We use databases of spectral irradiance measurements of outdoor downwelling light taken at regular intervals from dawn to dusk, at different locations in the northern hemisphere. Daily spectral data were collated by month. Estimates of the maximum velocity of daylight chromaticity changes for each month were obtained from smooth fits to the aggregated data, taking into account all possible CIELUV white points from the daylight locus. In all months and locations, the fastest CCT changes occur at sunset and sunrise, but speeds depend on day length and vary throughout the year. For a 2-year dataset from Granada, Spain (Hernández-Andrés et al., 2007), the largest velocity estimate is about 0.02 ΔEu*v*/s, occurring in January. Similar estimates (0.03 ΔEu*v*/s) are found for a 20-day dataset from Stanford, USA at the same latitude (37º) and month (DiCarlo & Wandell 2000). Note that here, concomitant changes in lux - which may be significant - are not considered. The maximum velocity of chromaticity change in daylight is much slower than the minimum illumination velocity discriminated by the observers in our previous experiments (0.5 ΔEu*v*/s), at least for these daylight measurements. These results suggest that outdoor CCT changes are too slow to be detected.