Right after this shot, I turned to the left and saw the possibility of making a paranoma, so I shot the city skyline. (Yes, I shot this first.)
When I got the prints, I could stitch the two together, but the exposure were different. Well, prints are never 100% repeatable. The scans will be accurate, I told myself.
But when I scan it in, the exposure were still different. Ah, I forgot to lock the exposure. So, I auto-exposed the other image, lock the exposure and then used it for this image. It's still wrong!
Could it be because the metering is different -- I might be using aperture priority mode. The large span of sky may fool the camera into underexposure. I checked, and the exposures are the same!
I can't solve this mystery. And I'm too lazy to correct the exposure manually. :-)