Title: Automatic Dynamic Relevance Determination (ADRD) for Gaussian Process with Two Dimensional Functional Input

Summary: Gaussian Process regression is a commonly used technique to study computer models. While using time-varying functional inputs with Gaussian process, Automatic Dynamic Relevance Determination (ADRD) offers a way to identify predictive relevance of the input over the index space using different weight functions. We extend this methodology to two dimensional functional input and demonstrate using Water Erosion Prediction Project (WEPP) data using multiple weight functions.