Abstract
Long-term Micro-Erosion Meter (MEM) datasets are commonly summarised using average downwearing rates, masking the hierarchical spatial and temporal variability they contain. We analyse a 13-year micro-erosion meter (MEM) monitoring dataset (2011–2024) from the tectonically stable Otway coast, southeastern Australia, using an integrated statistical framework combining robust linear mixed-effects modelling (RLMM), piecewise (broken-stick) mixed-effects analysis, and hierarchical centred autoregressive AR(1) modelling. The integrated modelling framework developed in this study explicitly resolves spatial hierarchy, non-linear temporal structure, and downwearing memory across nested platforms, MEM bolt sites, and point scales. RLMM results demonstrate strong scale dependence in downwearing variability: point-scale observations exhibit high variance and skewness, whereas platform-scale mean downwearing rates are comparatively stable. Relative to classical mixed-effects models, robust estimation substantially reduces residual variance and moderates slope estimates, indicating that extreme MEM values represent temporary spikes in downwearing rates rather than persistent geomorphic trends. Piecewise modelling identifies an early monitoring phase characterised by elevated downwearing rates, followed by a lower long-term downwearing rate, indicating a distinct transition between early and sustained monitoring phases. Hierarchical centred AR(1) modelling shows that downwearing rates are partly influenced by their previous values at the point (φ₁ = 0.325), site (φ₁ = 0.353), and platform (φ₁ = 0.382) scales, indicating that past downwearing conditions continue to influence present-day downwearing. Using the 2011–2024 MEM record, representing the contemporary monitoring period of the long-term Otway dataset, the hierarchical modelling framework reveals scale-dependent mean downwearing rates of 0.264–0.561 mm/yr, highlighting the value of the approach for resolving spatial and temporal structure. Together, these results demonstrate that MEM-derived downwearing rates provide a statistically robust, process-consistent framework for interpreting long-term shore platform evolution.