Interpretation of graphs is perhaps more a question relevant for a dedicated physics or math forum... You are right, however, that your received signal consists of your true signal (which can have contributions all over the frequency domain, not just low and middle), as well as some noise (which can also have contributions all over the frequency domain, although mostly its high frequency).
Your figure on the right confirms this. When you apply both filters, everything beyond 10 Hz goes close to 0; all high frequency signals are killed. However, you also seem to lose some of your signal between 1 and 10 Hz with the F=135 filter, and most of that same signal with your F=294 filter. This also comes into play on your original signal graph; around t=1400 ms ish, you have some modulation in your power, and your F=294 filter seems to "mean that out" so to speak, whereas the f=135 filter retains the "up down" look of your signal. I would be very hesistant before making any interpretation based on your green line - I would take the F=135 filter, or perhaps an even lower threshhold, maybe 100 ms, and work from there, unless you are very, very sure that the 1-10 Hz peaks are mostly noise.
