If the dynamics of a process is chaotic then uncertainty on the initial conditions typically grow exponentially, but if we have deterministic chaos the uncertainty can be reduced by measuring the initial state with high precision.
If external noise (randomness) is present the uncertainty cannot be reduced below a certain value by measuring the initial state.
For real world problems chaos and external noise are often present at the same time. Entropy is useful for measuring the uncertainty for both deterministic systems and systems with external randomness.
If the measurement precision is kept at a fixed level one cannot distinguish between deterministic chaos and external noise.