Causal Inference in GeoScience: From the Temporal to Spatial Dimensions

In past decades, increasing robust causal models were proposed, making causal inference under different scenarios and data limitations feasible. On one hand, these causal model are all based on time series data sources. On the other hand, in Earth Science, some variables, such as soil features and elevation, do not present a time series or the time series of these variables do not present sufficient temporal variations. In this case, traditional temporal causal models may fail to identify these clearly existing causalities in Earth Science.  To fill these gaps, here we show a Geographical Convergent Cross Mapping (GCCM) model for spatial causal inference with spatial cross-sectional data based cross-mapping prediction in reconstructed state space. Three typical cases, where clearly existing causations cannot be measured through temporal models, demonstrate that GCCM could detect weak-moderate causations when the correlation is not significant. And when the coupling between two variables is significant and strong, GCCM is advantageous in identifying the primary causation direction and better revealing the bidirectional asymmetric causation, overcoming the mirroring effect. The principle and some cases of GCCM are briefly introduced.