Visual albedo of biochar:
Verifying lambertian (cosine) sensitivity of a luxmeter sensor. Configuration
at 0.1 rad ‘angular height’ (84.3° from normal incidence).
Two details of measurement. The stick casting a shadow onto the sensor
is to be held just a bit shifted when the sensor is in sunshine, to block a
piece of the sky close to sun – this way, the bias is minimised:
Direct oblique sunshine, as apparent, is (18.1-7.8=10.3) klx.
A and an analogous measurement at normal incidence, using a hand to cast
the needed shadow:
Direct normal sunshine, as apparent, is (101.5-14.7=86.8) klx. In the preceding, oblique case at (π/2 - 0.1) rad, it should be 10× less, but it is not so little. We may guess that the sensor reports some 1.2× larger signal than it should, at this oblique direction.
Would it do any problem? For scenes of uniform luminance, the relative amount of light from such a direction, which is very far from a perpendicular one, is little anyway. The integral of effective space angle from 0 to x=π/2 - 0.1 is pi*(1-cos(2*x))/2 = ~3.110, just 1 % less then π. A 20 % error in the measurement of such extremely obliquely incident light can be mostly neglected. Of course, less oblique directions are more important and are to be inspected in a similar way.