<p>Ice shocks associated with the springtime thermal expansion of ice represent a manifestation of fracture processes in a complex nonlinear medium. In this study, we investigate temporal changes in statistical characteristics derived from ice deformation measurements preceding such events. The analysis demonstrates that the approach of ice shocks is accompanied by the formation of ordered linear features that constrain the evolution of a statistical functional constructed from deformation data. These features indicate a transition from predominantly stochastic behavior to the emergence of deterministic patterns in the system dynamics. A detailed examination of the temporal evolution of the statistical dependencies reveals several distinct groups of short-term precursors. The earliest indicators appear approximately one hour before an ice shock, followed by a second group emerging on the order of tens of minutes prior to the event. A further concentration of precursor signatures is observed within the final minutes preceding ice failure. Particular attention is given to the increasing topological similarity of the statistical patterns observed during these intervals, suggesting a progressive synchronization of deformation processes prior to fracture. The results support the applicability of statistical and topological analysis for identifying short-term precursors of ice shocks and highlight the potential of this approach for studying explosive-like processes in other complex natural systems.</p>