Most cameras are more or less calibrated to reproduce luminance correctly. Here, I want to derive the sun and sky radiance, calibrated in W/m2/sr, from multiple exposures of the sky. This complicates the issue because of spatial and temporal non-linearities.
What would be the best or most accurate methods to do this, considering that the direct, diffuse and global irradiance components are also measured with radiometers?
Chris, You could calibrate each of the channels as if they were reference cells. You would need to know the responsivity of each channels. You could use one Si reference cell, a light source, and spectroradiometer to calibrate each cell. You may need to restrict the field of view to get better data so there are not a lot of spectral variations across the field of view. Ideally you would then combine the 3 bands (tri-stimulus method) to give a single irradiance. This will not accurately capture the water bands but should do ok for the UV/Visible irradiance. There will be a spectral error as the spectral irradiance differs from the irradiance it was calibrated under. In principle if I went outside with a 5 degree filed of view for calibration and neglected temperature errors then I should be accurate (95% confidence limit) to 1% using a 0.35% absolute cavity radiometer to measure the total irradiance.
Thanks, Keith.
The current application is for a limited field of view (roughly 20°) centered on the sun, which should be easier to manage than fisheye lenses. Still, the problem is the likely variation in radiometric calibration due to the varying exposure times, which are themselves required to capture the whole dynamic range between dark sky and center of sun.
A recent paper (www.atmos-meas-tech.net/7/2137/2014/) presented a calibration method for a fisheye camera, but I wondered if something simpler, and hopefully not requiring sophisticated instrumentation (like reference cells, not available here), could be done.
I think it depends a lot on the camera you are using, though hopefully you have this covered already. Bear in mind that consumer cameras do a lot of in-camera color processing to make an image that will look good to the end-user, and the algorithms are generally proprietary, But there are cameras which can be put into a mode to output raw values. And as Keith said you'll have to search to find the responsivities for that camera's RGB sensors.
Chris, as raw method let me suggest thermal light source e.g. tiny 12V control light bulb from old car dashboard. You can put it in the edge of field of view of your camera. To use it as self calibration device it is necessary to know spectral response of your camera together with its optics in all 3 channels i.e. RGB – it is sophisticated but it is needed to do it once. (Or ask manufacturer) Then you can use the light bulb as thermal light source with well defined spectrum using Planck distribution. Total energy so the temperature of the filament can be calculated from voltage and current measurements – or find IR camera.
- Badges
- Science topic