Presenter/Author Information

Simon Hood
David Swayne

Keywords

pear, multi-objective optimization, model calibration, swat, sce

Start Date

1-7-2010 12:00 AM

Abstract

Recent work involving attempts to calibrate the Soil and Water Assessment Tool (SWAT) non-point source model has led to application of machine learning techniques (specifically the Apriori algorithm) to test runs of the model for particular watersheds. A new Parsimonious Explicit Apriori Reduction (PEAR) method for model calibration is evinced, and details of experiments demonstrating the improved efficiency and accuracy, as opposed to both the manual approach and a well-known Genetic Algorithm (GA), are outlined. The PEAR method overcomes difficulties intrinsic to three current classes of multi-objective optimization.

Share

COinS
 
Jul 1st, 12:00 AM

A Novel Model Calibration Technique Through Application of Machine Learning Association Rules

Recent work involving attempts to calibrate the Soil and Water Assessment Tool (SWAT) non-point source model has led to application of machine learning techniques (specifically the Apriori algorithm) to test runs of the model for particular watersheds. A new Parsimonious Explicit Apriori Reduction (PEAR) method for model calibration is evinced, and details of experiments demonstrating the improved efficiency and accuracy, as opposed to both the manual approach and a well-known Genetic Algorithm (GA), are outlined. The PEAR method overcomes difficulties intrinsic to three current classes of multi-objective optimization.