Figure 3
From: Predicting regional coastal sea level changes with machine learning
The regional modeled and predicted patterns are consistent with observed sea level variability. (A) Time series of the GP model (in blue) and observed sea level estimates (black) for the west Pacific Ocean region (shown in Fig. 2). (B) Same as in A but here the holdout method was used to predict sea level for the last 3 years of the record (in gray) by training the GP model from 1993 to 2015. (C) Sea level predictions for 1 (red), 2 (purple) and 3 (orange) years using the RNN method by training the observed sea level estimates (black) from 1993 to 2017, 2016 and 2015, respectively. The shaded envelops depict 95% prediction intervals. Sea level expressed in mm.
