Areal parameter estimates from multiple datasets

Abstract

A wide range of methods exist for interpolation between spatially distributed points drawn from a single population. Yet often multiple datasets are available with differing distribution, character and reliability. A simple scheme is introduced to allow the fusion of multiple datasets. Each dataset is assigned an a priori spatial influence zone around each point and a relative weight based on its physical character. The composite result at a specific location is a weighted combination of the spatial terms for all the available data points that make a significant contribution. The combination of multiple datasets is illustrated with the construction of a unified Moho surface in part of southern Australia from results exploiting a variety of different styles of analysis.

The transition from the Thomson Orogen to the North Australian Craton from seismic data

“…The Moho surface displayed in Figure 6 has been constructed using the procedure developed by Kennett (2019) in which the estimate of the local Moho at a spatial point is constructed from a sum of contributions from all available data points with allowance for weighting and the distance of these data points from the sample point. For different data sets we assign internal weights based on the quality of the result, e.g., for reflection picks we have used A = 0.9, B = 0.8, C = 0.7, D = 0.6.…”

“…We have followed Kennett (2019) to use an empirical relation, calibrated against receiver function results, to give an uncertainty 𝑒 in kilometres from a data weight 𝑤: 𝑒 = 0.8 + 6.0 * (1 − 𝑤). In Figure S4 we show the consistency and uncertainty estimates associated with Moho surface displayed in Figure 7.…”