Estimating aquifer recharge in Mission River watershed, Texas: model development and calibration using genetic algorithms |
| |
| Authors: | V Uddameri M Kuchanur |
| |
| Institution: | (1) Department of Environmental Engineering, MSC 213, Texas A&M University-Kingsville, Kingsville, TX 78363, USA |
| |
| Abstract: | Soil moisture balance studies provide a convenient approach to estimate aquifer recharge when only limited site-specific data
are available. A monthly mass-balance approach has been utilized in this study to estimate recharge in a small watershed in
the coastal bend of South Texas. The developed lumped parameter model employs four adjustable parameters to calibrate model
predicted stream runoff to observations at a gaging station. A new procedure was developed to correctly capture the intermittent
nature of rainfall. The total monthly rainfall was assigned to a single-equivalent storm whose duration was obtained via calibration.
A total of four calibrations were carried out using an evolutionary computing technique called genetic algorithms as well
as the conventional gradient descent (GD) technique. Ordinary least squares and the heteroscedastic maximum likelihood error
(HMLE) based objective functions were evaluated as part of this study as well. While the genetic algorithm based calibrations
were relatively better in capturing the peak runoff events, the GD based calibration did slightly better in capturing the
low flow events. Treating the Box-Cox exponent in the HMLE function as a calibration parameter did not yield better estimates
and the study corroborates the suggestion made in the literature of fixing this exponent at 0.3. The model outputs were compared
against available information and results indicate that the developed modeling approach provides a conservative estimate of
recharge. |
| |
| Keywords: | Water balance Soil moisture Vadose zone Aquifer recharge Texas USA |
| 本文献已被 SpringerLink 等数据库收录！ |
|
