Impact of Data Assimilation Frequency and Observation Location in Thermal Effluent Modeling for Coastal Waters
Abstract This study investigates the application of data assimilation (DA) using the Ensemble Kalman Filter (EnKF) to address the uncertainties associated with modeling the thermal effluents discharged from power and desalination plants into a shallow, tidal bay. A two‐dimensional hydrodynamic model...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
American Geophysical Union (AGU)
2025-07-01
|
| Series: | Earth and Space Science |
| Subjects: | |
| Online Access: | https://doi.org/10.1029/2024EA004099 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract This study investigates the application of data assimilation (DA) using the Ensemble Kalman Filter (EnKF) to address the uncertainties associated with modeling the thermal effluents discharged from power and desalination plants into a shallow, tidal bay. A two‐dimensional hydrodynamic model of Sulaibikhat Bay (SB), Kuwait, was developed using the Delft3D Flexible Mesh Suite to simulate the transport and dispersion of a thermal plume under the forces of predominantly semidiurnal tides. A series of observing system simulation experiments were conducted using the OpenDA toolbox to identify the optimal observation location and DA frequency. Significant reductions in temperature prediction errors were achieved using the EnKF for state estimation. The optimal DA frequency was characterized by its ability to retain the analysis adjustments in the system while maintaining computational efficiency. Relative to the dynamics of SB, the optimal frequency was found to be hourly. An excessive rate of DA was found to cause filter divergence, where forecast error is falsely underestimated due to diminishing ensemble variance. This leads the filter to ignore the information provided by the observations. In contrast, a sparse rate of DA was found to cause the model to revert to its pre‐assimilative state. The optimal station locations were identified using the ensemble‐based targeted observation method, based on their ability to maximize the reduction of the analysis error variance. The optimal locations were attributed to having a balance between exhibiting strong covariances with the other state elements while experiencing local variance exceeding that of the prescribed observation error variance. |
|---|---|
| ISSN: | 2333-5084 |