Overland flow: interfacing models with measurements
Index words: overland flow, catchment scale, system identification, ensemble simulations.<br/><br/>This study presents new techniques to identify scale-dependent overland flow models and use these for ensemble-based predictions. The techniques are developed on the basis of overland flow, rain, discharge, soil, vegetation and terrain observations that were collected over a three year period in two tropical catchments. The merits of the identification technique are its robustness with regard to unknown errors, the ability to adjust model resolution in response to data availability, and to interpret the entities of the identified model structures physically. Compared to a static regression model and a dynamic distributed model the predictive performance of the scale-dependent overland flow models is good, especially when using model ensembles. Further analysis of the scale-dependent models shows that rainfall largely determines overland flow when modelled at coarse resolutions, whereas soil moisture drives overland flow when defined at fine resolutions. Interestingly, the number of model parameters remains constant over the different resolutions. The use of the scale-dependent models for predictive purposes is demonstrated by applying Tikhonov regularization for recursive state as well as parameter estimation.
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Doctoral thesis biblioteca |
| Language: | English |
| Subjects: | hydraulic conductivity, models, runoff, soil water movement, bodemwaterbeweging, hydraulisch geleidingsvermogen, modellen, oppervlakkige afvoer, |
| Online Access: | https://research.wur.nl/en/publications/overland-flow-interfacing-models-with-measurements |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Index words: overland flow, catchment scale, system identification, ensemble simulations.<br/><br/>This study presents new techniques to identify scale-dependent overland flow models and use these for ensemble-based predictions. The techniques are developed on the basis of overland flow, rain, discharge, soil, vegetation and terrain observations that were collected over a three year period in two tropical catchments. The merits of the identification technique are its robustness with regard to unknown errors, the ability to adjust model resolution in response to data availability, and to interpret the entities of the identified model structures physically. Compared to a static regression model and a dynamic distributed model the predictive performance of the scale-dependent overland flow models is good, especially when using model ensembles. Further analysis of the scale-dependent models shows that rainfall largely determines overland flow when modelled at coarse resolutions, whereas soil moisture drives overland flow when defined at fine resolutions. Interestingly, the number of model parameters remains constant over the different resolutions. The use of the scale-dependent models for predictive purposes is demonstrated by applying Tikhonov regularization for recursive state as well as parameter estimation. |
|---|