On the robustness of elastic thickness estimates obtained using the coherence method

In order to uniquely determine the elastic thickness of the lithosphere, T e , from gravity and topography data, the coherence method explicitly assumes that surface and subsurface loads are statistically uncorrelated. In many realistic cases (e.g., mountain building) this assumption is likely to be violated. We present the results of Monte Carlo simulations designed to measure the bias and variance of T e estimates when the degree of correlation, R , between fractal surface and subsurface loads is not null. In this study, R denotes the linear correlation coefficient (Pearson's coefficient), and surface and subsurface loads are self‐affine (fractal). Recording to our results, when there is no correlation between the loads ( R = 0), the estimatedT e ˆis in excellent agreement with the simulated T e although its accuracy decreases for increasing plate stiffness. We also find that there is a strong likelihood of an upward bias inT e ˆestimates if the dimensions of the study area are not adequate to fully resolve the longest wavelengths. This might partially explain why T e values for previous continental studies obtained using the coherence method are often higher than T e values obtained using conventional forward modeling techniques. As the degree of correlation increases, we observe a clear downward bias in the estimatedT e ˆ . When surface and subsurface loads are correlated by only a small amount, the fit between the computer‐simulated and predicted coherences is still good for the longest wavelengths within the diagnostic wavebands. In light of these limitations, we use the coherence method to reassess T e estimates for the Alps. Our results indicate that the flexural behavior of the lithosphere to loading does not change significantly along the fold/thrust belt. The similarity between our T e results for the western (30–39 km) and eastern Alps (33–40 km) does not support the hypothesis that T e and the horizontal radius of curvature of mountain belts are correlated.