Look at a scene and notice whether is has a wide range between bright and dark or a narrow range. If a wide range, it is likely the metering in matrix will be a very good job of finding the mid point that is how you expect. If a narrow range, like these scenes, which are on the brighter end of the spectrum...bright cloudy sky and light colored buildings, the meter assumes the mid point of bright and darkest is supposed to look grey in black/white balance. So seeing that scene before even looking through the camera VF, one would assume the image would exposed as mostly grey brightness level. If there was a large dark object in the frame, the range would be wider between the bright sky and the darkest elements so exposure would probably be exactly how you expected.
Since there is no dark portion of the scene we know it is going to be under exposed to make the bright areas closer to that 18% grey. Dialing in 1.5 stops of positive exposure compensation before taking the shot would have made it look like you wanted it.
If you were filling the frame with a black cat, the range of brightness would be limited and low but the camera thinks the mid point between the brightest and darkest is 18% grey, the black cat would be overexposed and dialing in 1-2 stops of negative exposure compensation would make the scene look as you intended. This difference between expected and and results happens when the brightness range is narrow, and when wider, the meter will be much closer to what you see with your eyes. The camera has an Ez-Compensation function that allows the command wheels on the right be used for quick exposure compensation. You generally do not need it with most scenes that have both bright and dark elements.
