Evaluation of Dynamically Dimensioned Search Algorithm for Optimizing SWAT by Altering Sampling Distributions and Searching Range
Academic Article
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
2016 American Water Resources Association. The primary advantage of Dynamically Dimensioned Search (DDS) algorithm is that it outperforms other optimization techniques in both convergence speed and searching ability for parameter sets that satisfy statistical guidelines while requiring only one algorithm parameter (perturbation factor) in the optimization process. Conventionally, a default value of 0.2 is used as the perturbation factor, where a normal distribution is applied with mean sampling distribution of zero and variance of one. However, the perturbation factor sensitivity to the performance of DDS for watershed modeling is still unknown. The fixed-form sampling distribution may result in finding parameters at the local scale rather than global in the sampling space. In this study, the efficiency of DDS was evaluated by altering the perturbation factor (from 0.05 to 1.00) and the selection of sampling distribution (normal and uniform) on hydrologic and water quality predictions in a lowland agricultural watershed in Texas, United States. Results show that the use of altered perturbation factor may cause variations in convergence speed or the ability to find better solutions. In addition, DDS results were found to be very sensitive to sampling distribution selections, where DDS-N (normal distribution) outperformed DDS-U (uniform distribution) in all case scenarios. The choice of sampling distributions could be the potential major factor to be attributed for the performance of auto-calibration techniques for watershed simulation models.
