The phenomenon of predictability worsening is studied, which is characteristic for chaoticdynamics, reconstructed from incomplete set of observational data. It is pointed out that inconditions of data deficiency, when there are fewer observables than independent vari-ables, reconstruction procedure inevitably has to deal with additional differentiations ofnoisy observables, which is the main reason for the phenomenon of predictability worsen-ing to take place, especially in the presence of short-correlated noise. The phenomenon of predictability worsening is illustrated with a theoretical analysis andnumerical simulations for the third order system (R_ssler attractor), which can be recon-structed using the least squares method on the basis of three, two and one noisy observ-ables. For all the three cases the admissible noise intensity is estimated, which providesacceptable quality of prediction. It is shown that a deficiency of every observable is respon-sible for a significant (up to 20 ÷ 100 times) decrease of the admissible noise. Numericalresults are in a satisfactory agreement with theoretical expectations.
展开▼
