stocpotato correctly pointed out that my dimensions were not stated correctly. I had forgotten the 0.2 m2 surface area of ROMIT. And if that is taken into account, the deviation after 1000 seconds is 9 cm or after a month close to a thousand km, if the sunlight is fully absorbed. Since most of it is reflected, the actual deviation must be higher than that. So the point that sunlight cannot be ignored in the calculations, still stand. Tor