Figure 3.

Visualising the spectral diet over a day.

The spectral variation of light exposure over one work day (‘spectral diet’) is captured with an integrated wearable sensor. The experienced ‘spectral diet’ is visualised by plotting the normalised spectral information over time (a), where variation in the relative power is shown with a gradient colour map. Accounting for statistical regularities, we conceptually propose to reduce the dimensionality of the spectral diet to a small group of representative spectra selected by running a k-means clustering algorithm on the set of all measured spectra (b). This reduction shows that only a small set of spectra may effectively be able to describe the majority of the spectral information reached at the eye. The identification of common spectral ‘species’ might mean that we can use statistical regularities in the built environment to constrain the complexity of ‘decoding’ the spectral diet of humans. While biological parameters may reflect the relative effect of light exposure over time (c), the illuminance conditions shown on a logarithmic scale highlight the variation in high intensity daylight and lower intensity electric lighting (d). The device used to collect spectral data (a) includes a novel small-footprint array-based spectrometer developed by nanoLambda (Korea), which allows for variable sampling rates, data storage on-board or via bluetooth (approximate operating range: 5–40,000 l×).