Colour patterns from the reflection of my window 

The first picture is that of my window while the second is its reflection as seen from the floor. The pattern depends on the angle. That is when I move a little closer then look at the reflection the left part of window becomes transparent and the pattern is only seen at the middle and the end.
What causes this?
I don't think it can be thin-film interference from any thin layer formed at the surface of my floor since the pattern is uniform while the floor cannot be so smooth.
Also if this was caused by dispersion then the white light from my led should also get dispersed and I don't think that the dispersion would depend on the angle.
These are my attempts to understand this. Any input is appreciated.
Some new observations:
1)The pattern is only visible when the window is closed. If I keep a mirror on the floor where the pattern is formed , the reflection from mirror causes no pattern while the rest shows the same pattern.
2)The pattern is most prominent at the evening when the Sun is almost below the horizon (sunset). I have completely ruled out thin-film interference with this new data.
My best guess now is that my window mirror absorbs some colors and this is enhanced by the reflection from the floor.
Related to Alphy's answer

*

*How does the thicker thinner argument reconcile with the uniform colours  and almost constant widths of each colour at a slanted angle? Also the colours get repeated  after some interval  which would imply that thickness decreases then increases or that the angle somehow (still confused about the angle argument) to give thaf red which seems unlikely.

*You have mentioned that reflected light enhances the effect (which I also think) but putting a mirror where image is formed does not give any colour. So why would reflection from floor only do this and not the mirror?
3)"The glass changes the angle (and it does so differently for different wave-lengths) at different points, therefore, light gets censored differently at different points" yet the colours have a width and are uniform. This would imply that the change in angle at different points etc etc would have to happen in a very special way to give this pattern.



 A: What is happening, in my opinion, is the result of two effects:
First one is to notice that the glass of your window is most probably not uniform. Meaning, as a guess, it is thinner on the top and gradually gets thicker to the bottom, but in a very slight manner that only light feels it. What does this cause? Well if the glass also has two different refraction indices, say $n_v$ for vertical refraction index and $n_h$ for horizontal refraction index, then the polarization of the light traveling through the glass will rotate due to this non-isometry $n_h \neq n_v$. But remember that the glass at some point is thicker than the others; which leads to the fact that the polarization rotates more at those points that are thicker.
The second is to note that the incoming light is probably polarized to some extent since, from the looks of it, it is reflected from the wall in front of your window. In general reflected light gets polarized and this effect is more intense for certain reflection angles. So you have a polarized light which rotates differently at different points and then gets one more time reflected from the ground. This second reflection also polarizes the light one more time and cuts out those lights that are opposing its direction of polarization.
Therefore, essentially, some of the reflected light from the ground gets censored (by the procedure of polarization). Which light is censored depends on the angle of polarization of the light. The glass changes the angle (and it does so differently for different wave-lengths) at different points, therefore, light gets censored differently at different points and for different colors; leading to the pattern you are seeing.
Now you might ask:

*

*How does the thicker thinner argument reconcile with the uniform colours and almost constant widths of each colour at a slanted angle? Also the colours get repeated after some interval which would imply that thickness decreases then increases or that the angle somehow (still confused about the angle argument) to give that red which seems unlikely.


*You have mentioned that reflected light enhances the effect (which I also think) but putting a mirror where image is formed does not give any colour. So why would reflection from floor only do this and not the mirror?


*"The glass changes the angle (and it does so differently for different wave-lengths) at different points, therefore, light gets censored differently at different points" yet the colours have a width and are uniform. This would imply that the change in angle at different points etc etc would have to happen in a very special way to give this pattern.
To answer these queries it is simpler to imaging that the incident light includes only one frequency (let's say red) of the whole visible spectrum.

*

*If the thickness of the glass gradually increases from top to bottom, then our red polarized light gradually rotates from top to bottom. Of course, the transformation of rotation is periodic, therefore after a while the direction of polarization comes back to its initial state. This then leads to having an image constructed by ordered bands of dark and light red.


*The reflected light is polarized to some extent but there is an optimum angle for polarization that is different in different material. I suspect that if you look at the mirror while holding it with a different angle than the floor (probably almost perpendicular to the incident light) then you might be able to see the same phenomenon.


*It is kind of a special way, since it is periodic.
A: Your colour effect is somewhat similar to what can be seen through the windows of some trains. A part of the blue sky is a source of polarised light. When viewing that part of the sky through a birefringent window and a linear polarisation analyser, it is possible to see the colour pattern. It is more likely (and cheaper) a window is birefringent due to a polymer plastic film, than due to the glass.

The sky seen through a train window, with a linear polariser on the camera. There is no film in the corner.
A: Thanks for making (and showing) extra pictures.
In the second extra picture, I can see faintly the described effect in the reflection by the floor but not in the reflection by the mirror. Just look on your screen at a small angle from below, which enhances the effect. So even if the window is open, the effect is visible, so we can eliminate the effect of the window. In the first picture (is it taken on a cloudy day?), no pattern is visible indeed. But you took the picture much closer to the window (so changing the angle). As the effect I describe is angle-dependent this could be possible (you write: That is when I move a little closer then look at the reflection the left part of the window becomes transparent and the pattern is only seen in the middle and the end).
Plate glass has only one refractive index (independent of the orientation see here: 1.52). Its thickness may vary but this gives no visible interference effects because of the thickness of the plate.
So we are left with only one possibility: The floor is responsible. If you can see this effect with the naked eye, the cause can be the same as the cause of rainbow patterns of oil on water. But since you ruled out this cause because of the roughness of the floor without a layer, the cause must be due to the effect of the floor on all polarizations.
This means that all polarizations are reflected in slightly different ways (just as in the water-oil case).
