Improving hydrologic modeling of runoff processes using data-driven models
| dc.contributor.author | Han, Heechan, author | |
| dc.contributor.author | Morrison, Ryan, advisor | |
| dc.contributor.author | Grigg, Neil S., committee member | |
| dc.contributor.author | Bailey, Ryan T., committee member | |
| dc.contributor.author | Kampf, Stephanie, committee member | |
| dc.date.accessioned | 2021-06-07T10:21:01Z | |
| dc.date.available | 2021-06-07T10:21:01Z | |
| dc.date.issued | 2021 | |
| dc.description.abstract | Accurate rainfall–runoff simulation is essential for responding to natural disasters, such as floods and droughts, and for proper water resources management in a wide variety of fields, including hydrology, agriculture, and environmental studies. A hydrologic model aims to analyze the nonlinear and complex relationship between rainfall and runoff based on empirical equations and multiple parameters. To obtain reliable results of runoff simulations, it is necessary to consider three tasks, namely, reasonably diagnosing the modeling performance, managing the uncertainties in the modeling outcome, and simulating runoff considering various conditions. Recently, with the advancement of computing systems, technology, resources, and information, data-driven models are widely used in various fields such as language translation, image classification, and time-series analysis. In addition, as spatial and temporal resolutions of observations are improved, the applicability of data-driven models, which require massive amounts of datasets, is rapidly increasing. In hydrology, rainfall–runoff simulation requires various datasets including meteorological, topographical, and soil properties with multiple time steps from sub-hourly to monthly. This research investigates whether data-driven approaches can be effectively applied for runoff analysis. In particular, this research aims to explore if data-driven models can 1) reasonably evaluate hydrologic models, 2) improve the modeling performance, and 3) predict hourly runoff using distributed forcing datasets. The details of these three research aspects are as follows: First, this research developed a hydrologic assessment tool using a hybrid framework, which combines two data-driven models, to evaluate the performance of a hydrologic model for runoff simulation. The National Water Model, which is a fully distributed hydrologic model, was used as the physical-based model. The developed assessment tool aims to provide easy-to-understand performance ratings for the simulated hydrograph components, namely, the rising and recession limbs, as well as for the entire hydrograph, against observed runoff data. In this research, four performance ratings were used. This is the first research that tries to apply data-driven models for evaluating the performance of the National Water Model and the results are expected to reasonably diagnose the model's ability for runoff simulations based on a short-term time step. Second, correction of errors inherent in the predicted runoff is essential for efficient water management. Hydrologic models include various parameters that cannot be measured directly, but they can be adjusted to improve the predictive performance. However, even a calibrated model still has obvious errors in predicting runoff. In this research, a data-driven model was applied to correct errors in the predicted runoff from the National Water Model and improve its predictive performance. The proposed method uses historic errors in runoff to predict new errors as a post-processor. This research shows that data-driven models, which can build algorithms based on the relationships between datasets, have strong potential for correcting errors and improving the predictive performance of hydrologic models. Finally, to simulate rainfall-runoff accurately, it is essential to consider various factors such as precipitation, soil property, and runoff coming from upstream regions. With improvements in observation systems and resources, various types of forcing datasets, including remote-sensing based data and data-assimilation system products, are available for hydrologic analysis. In this research, various data-driven models with distributed forcing datasets were applied to perform hourly runoff predictions. The forcing datasets included different hydrologic factors such as soil moisture, precipitation, land surface temperature, and base flow, which were obtained from a data assimilation system. The predicted results were evaluated in terms of seasonal and event-based performances and compared with those of the National Water Model. The results demonstrated that data-driven models for hourly runoff forecasting are effective and useful for short-term runoff prediction and developing flood warning system during wet season. | |
| dc.format.medium | born digital | |
| dc.format.medium | doctoral dissertations | |
| dc.identifier | Han_colostate_0053A_16465.pdf | |
| dc.identifier.uri | https://hdl.handle.net/10217/232583 | |
| dc.language | English | |
| dc.language.iso | eng | |
| dc.publisher | Colorado State University. Libraries | |
| dc.relation.ispartof | 2020- | |
| dc.rights | Copyright and other restrictions may apply. User is responsible for compliance with all applicable laws. For information about copyright law, please see https://libguides.colostate.edu/copyright. | |
| dc.subject | deep learning | |
| dc.subject | machine learning | |
| dc.subject | data driven model | |
| dc.subject | National Water Model | |
| dc.subject | hydrologic modeling | |
| dc.title | Improving hydrologic modeling of runoff processes using data-driven models | |
| dc.type | Text | |
| dcterms.rights.dpla | This Item is protected by copyright and/or related rights (https://rightsstatements.org/vocab/InC/1.0/). You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s). | |
| thesis.degree.discipline | Civil and Environmental Engineering | |
| thesis.degree.grantor | Colorado State University | |
| thesis.degree.level | Doctoral | |
| thesis.degree.name | Doctor of Philosophy (Ph.D.) |
Files
Original bundle
1 - 1 of 1
Loading...
- Name:
- Han_colostate_0053A_16465.pdf
- Size:
- 4.52 MB
- Format:
- Adobe Portable Document Format