Noise is Beautiful, or Scales, or Bugs

All of these are interesting, and you probably
won't go wrong on any.  Bugs attract me more
for some reason, but maybe that's because
it's spring and bugs will take over soon anyway.

On Noise:
There are some classic examples of Spectral
Modeling from Xavier Serra's thesis and work
thereafter, wherein he sucks out the deterministic
(non-noise, sinusoidal) sounds, leaving only 
the noise.  Then he does some cool transformations.
I'll try to put up a directory of these to inspire you.
You might look into TAPESTREA, 'cause that's a good
tool for doing this type of work.

On Scales:
Might find a set of "instruments" that have interesting
non-harmonic spectra (like water glasses, or hunks of
pipe, or something semi-pitchy but non-harmonic), analyze
them, and devise a scale that compliments and emphasizes
their unique properties.  Roger Shepard and Carol Krumhansl
found that Javanese Gamelan scales are totally strange, 
unless you consider the spectra of the bells, then everything
makes much more sense.

On Bugs:
Your plan sounds great.  Pick a bug/critter type and 
sonically dissect them.

Forge Ahead!!