I want to preface this by saying that I could be totally wrong and invite anyone to correct me as this is just my best guess from one internet academic to another. But what i think is going on is this:
Any time digital audio exists it is created with both a sample rate and bit depth. The sample rate controls how high a frequency is accurately produced (how far to the right on your spectrogram) and the bit depth controls the level of detail in the sound.
Now this is where i might be a bit off but a common misconception is that the bit depth controls the accuracy of the actual waves reproduction. Aka the higher the bit depth, the less jagged the wave is. But because the sampling theorem states that any sampled wave can result in only one possible wave shape (along with skipping a lot of technical explanation and such stuff) the bit depth can basically be summed up as controlling the noise floor of the sound.
I think that noise is coming from the internal bit depth of Operator. I also think it starts to slightly ramp up in the higher frequencies of the spectrogram because aliasing, which is noise created by the sample rate of the audio. You should be able to compare this by looking at the noise floor that is created when you play the 8 or 16 bit sine wave options in Operators waveshape selector.
Two other things to note is that a noise floor of -150Db is super quiet and you would never be able to hear it next to your sound so its nothing to worry about. Also in operator you can change the amount of aliasing noise created by messing with the tone knob next to the master volume control (although you won't be able to see its effect on such low volume high frequencies).