Why is "a very silent room" louder than "leaf noise" / "breathing noise"? I recently was at night in my bed room and used the Schallmessung Android App (on a Samsung S7). It showed 20 dB while laying on some furniture. I could not hear anything; to me, the room was completely silent.
Now I looked at some examples of noise levels on the German Wikipedia. Two of them confused me a lot:


*

*Sehr ruhiges Zimmer (very silent room): 20-30dB

*Blätterrauschen, ruhiges Atmen (leaf noise, breathing noise): 10dB


Why is a "very silent room" (I held my breath, no AC, at night, windows closed) where I can't hear anything louder than "leaf noise"? How accurate can I expect the results of the app to be?
 A: You should not trust your phone's noise measurement at such low sound levels. The electronics in the microphone generate a significant amount of electrical noise which becomes part of the microphone signal that your phone receives. This "self noise" creates a noise floor for your measurements, which means that:


*

*If the sound in the room is much higher than the self noise, the self noise is negligible and the measurement is fairly trustworthy if the microphone is calibrated and has a reasonably flat frequency response, as long as the sound is not strong enough to overdrive the microphone.

*If the sound in the room is lower than the self noise, the self noise dominates and your measurement does not tell you anything about the noise in the room.
I worked on an iOS app that included a sound level meter a few years ago, and this phenomenon was very noticeable. I don't remember where the noise floor was on the devices we tested, but it seems like the noise floor on your phone is around 20 dB, which is honestly pretty good.
A: A silent room is not perfectly silent, but contains sound from ventilation sources and vibration sources inside the building plus outside sounds being transmitted into the room through its walls and windows. These sources are accounted for in the (textbook) citation of the sound level inside a "typically quiet" room. 
A: I absolutely agree with Erlend's answer (RE the noise floor of a telephone application.) I also agree that a noise floor of 20 dB is remarkably low. The noise floor on my professional sound level meters are 14 dB and 5.5 dB, respectively. The latter is a special low-noise system.

However, here's an attempt to answer your question about why a reference may list "very silent room" as more energetic than "leaf noise". Outdoor environments represent a closer approximation of free field conditions, where sounds propagate outward without reflection.
Enclosed rooms are usually representative of the reverberant field, with reflections off walls and objects in the room. Acoustic energy stays in the room longer and results in higher values of measured pressure for the same source power.

For example: when I stop typing, one-second equivalent sound pressure levels ($L_{Aeq, 1s}$) in the center of this room are between 30 - 40 dB. Most of the energy is coming from my computer's fan and the ballast of the fluorescent light. Outside on a still day these sources would contribute far less energy to a measurement at their same distance.
This may provide an explanation for why a handbook would list even a "very silent" room as being more energetic than an outdoor environment with some motion. 
Frankly, I think these kinds of generalizations are far too broad, so I try to avoid them in my work. The prediction of ambient level is dependent on both the form and material structure of a space as well as the sources within/nearby. The sheer variety of combinations possible suggests that measurements are highly preferable to reliance on listed values. This includes an awareness of noise floor limitations, as Erlend's answer emphasizes.
A: They are most likely talking about just the "leaf noise" by itself. If you were in your silent room with additional "leaf noise" then your app would obviously record a higher decibel reading (assuming it is sensitive enough).
