In particular, Evans and Coombe (1959) estimated sunlight penetration through forest canopy openings by overlaying diagrams of the sun track on hemispherical photographs.
Later, Margaret Anderson (1964, 1971) provided a thorough theoretical treatment for calculating the transmission of direct and diffuse components of solar radiation through canopy openings using hemispherical photographs.
At that time hemispherical photograph analysis required tedious manual scoring of overlays of sky quadrants and the track of the sun.
With the advent of personal computers, researchers developed digital techniques for rapid analysis of hemispherical photographs (Chazdon and Field 1987, Rich 1988, 1989, 1990, Becker et al. 1989).
Various commercial software programs have become available for hemispherical photograph analysis, and the technique has been applied for diverse uses in ecology, meteorology, forestry, and agriculture.
Hemispherical photography has been used successfully in a broad range of applications involving microsite characterization and estimation of the solar radiation near the ground and beneath plant canopies.
These calculations require theoretical or empirical distributions of direct and diffuse radiation in the open, without canopy or other sky obstruction.
Usually calculations are performed for either photosynthetically active radiation (400-700 nanometers) or insolation integrated over all wavelengths, measured in kilowatt-hours per square meter (kW h/m2).
Upward-looking hemispherical photographs are typically acquired under uniform sky lighting, early or late in the day or under overcast conditions.
Recently most photographs are acquired using digital cameras (e.g., Kodak DCS Pro 14nx with a Nikkor 8mm fisheye lens).
Photograph classification involves determining which image pixels represent visible (non-obscured) versus non-visible (obscured) sky directions.
Recently advances have been made in developing automatic threshold algorithms, however more work is still needed before these are fully reliable.