A non-parametric regression model means that we can build the model from the observation data without parameters, it does not depend on the distribution in the dataset. The parametric model indicates the parameters which found in the model, for e.g the linear regression, you have to find some parameters such as beta0, beta1, .. using a square loss function to minimize them.
I am not familiar with DSAS, you can clarify your problem and I may find a solution for you.
Prediction of shoreline change can be predicted using two ways:
1. Using numerical model. You can use GENESIS numerical model to predict shoreline change under many scenarios, such as narurally, or using coastal structures (breakwater, revetment, etc.
2. Using DSAS. Using DSAS you can analysis statistically the rate of shoreline change based on shoreline hystorical data.