Using the Bayesian Statistical Framework for Testing Paleomagnetic Reconstructions
PI: Dr. Joseph Meert
University of Florida Department of Geological Sciences
1/2023 - 12/2023
​​
The theory of plate tectonics states that continents are in constant motion such that the geography of the Earth has changed over time. Paleogeography can be examined in deep-time (>600 million years ago) using a combination of paleomagnetism and geochronology. To decipher past geography, we need to know where the continents were located and when they occupied that location. There are errors associated with each measurement and are described using Fisherian statistics. The most common is called alpha-95, which describes an angular cone of 95% confidence about the mean direction. To reconstruct past geometries of continents, paleomagnetists examine similar age rocks from one or more continents. If those rocks yield a similar latitude, then the continents could be rotated together and examined to test whether the geological links were reasonable. More robust statistical methods are being developed for comparison which consider errors in age and position. Bayesian statistics allows us to test for overlap as a continuous probability from 0% (not similar) to 100% (identical). This project focused on using Bayesian probabilities to test ‘goodness of fit’ and a variety of proposed paleogeographies from 1000-1800 million years ago using these new methods along with other new tools being developed.
Figure 1. Coeval data from South China at 824 Ma (Yanbian Dykes-A=YA pole) and 770 Ma (Yangtze dykes=YD) can be used to reconstruct India and South China. The fit between the two pairs of similar age poles from both cratons is excellent (HD-YA Bayes error=.482 and MIS-YD Bayes error=.384; Meert and Santosh, 2022). If we
assume that South China and India were fellow travelers during the 825-770 Ma interval, then we can use other poles from South China to evaluate the drift rate of both cratons. We perform a bootstrap calculation for pole positions and Monte Carlo age simulations to evaluate apparent polar wander (APW) rates and latitudinal drift rates.
Figure 2. (a) Simulated Poles for Yangtze Block and India rotated into Indian coordinates. (b) Paleolatitude plot using simulated poles in (a). Rates of motion are shown as APW (Apparent polar wander) or latitudinal rates with associate confident intervals. When standard deviations are large, we also show median values.
Figure 3. Continental reconstruction of India and South China at 770 Ma and 820 Ma.