A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
From: Marcel Tschudin
Date: 2012 May 30, 18:18 +0300
I understand that the reflected angle is constant. The problem is your calibration (pixels per degree) which relate to the angle in the image without Bris. When you measure in the photo the pixels you get your altitude (with refraction related to the image angle). When you add now to this angle the the constant angle of the Bris the refraction difference between smaller and larger altitude remains unconsidered. Try to see in your various Bris observations whether you obtain by systematically adding or subtracting this refraction difference an improvement. This should really lead to an improvement.
Here how you should proceed:
1) Photo and calibration: Measure the pixels and convert to altitude = Alt1
2) Add to Alt1 the Bris constant angle = Alt2
3) Calculate the refraction for the two altitudes: Ref(Alt1) and Ref(Alt2)
4) Calculate the difference of refraction: DiffRef = Ref(Alt1) - Ref(Alt2)
5) Correct Alt2: Alt2corr = Alt2 +(or -) DiffRef (check whether you need the + or - sign)
Alt2corr is expected to be a better estimate than Alt2.
It is not as complicated as you think. The reflected Sun image is a constant angle that repeats exactly and requires only one single calibration to determine and is transferable to any lens or camera unchanged or simply used without a camera and timed as if a sunset. The loss of precision is in finding the vertical to the horizon since the Sun isn't near the horizon to be rocked back and forth for the low point. There is in addition a loss of image sharpness from added glass reflections that disrupt the symmetry of the Sun's reflected image.
[NavList] Re: Camera FAR 2nd Ed.
From: Marcel Tschudin
Date: 30 May 2012 14:08
Regarding the accuracy you attain with your various FAR systems: Could it be that the loss of accuracy results from an error in not considering or wrongly considering differential refraction? The calibration of your lens considers refraction between actual altitude and apparent horizon. The reflection of the Bris moves the image to an altitude where the original calibration (refraction) is no longer true. May be one can obtain a good approximation with the original curve by adding or subtracting the difference in refraction. Imagine the sun being at 50* altitude being refracted by Ref(50). This image is now shifted down by e.g. 28* 46' producing an image at 21* 14'. Your calibration curve considers refraction between this altitude and the apparent horizon, but not between 50* and the horizon. As a gross approximation you may try to correct this (by adding or subtracting ?) with the difference between Ref(50) and Ref(21* 14'). It is however my understanding that in order to do it correct requires to derive separate calibration curves for each of the two Bris images.