Complex system approach to interpretation of monitoring time series: two case histories from NW Bohemia

Abstract

Ever wider implementation of information technologies is flooding us by monitoring data. To an efficient risk management, those data have to be processed and assessed in the same rate as they are recorded and transported. Paper demonstrates some methods dealing with intrinsic, nonlinear dynamics of slope system for computerized safety assessment of monitoring time series, their modeling and early warning launching. Analysis and modeling of phase changes – i.e. specific transient states between different developmental stages of dynamics of unstable slope systems, enabled to fix new types of precursors for rock fall warning and to enhance time prediction of rock fall occurrence. Mathematically well based, novel numerical and topological methods from the toolbox of complex system theory were successfully implemented to that challenge. Their pattern-recognition ability, i.e., diagnostic sensitivity, and more realistic results of modeling of time series relevant patterns have reached beyond ranges of regularly used—both the idea- and the data-driven—methods. Moreover, results yielded by phase space analyses are in good agreement with the ones by numerical fractal analyses. Obtained results have strengthened the primary, mainly theoretically based hypothesis; the dynamics of an unstable rock slope has to be considered as behavior of nonlinear, dissipative, non-equilibrium, self-organizing complex system. Correspondingly to the theoretical assumptions, two qualitatively different types of slope system dynamics—near to equilibrium and far from equilibrium states, were identified. All field monitoring data used originated in practice of highly automated, integrated IT system of monitoring assessment, and management of rock fall hazard at sandstone rock walls in the NW Bohemia.