Abstract In this proposed research, we will model peak flowrates in Nebraska streams using new high-resolution datasets and a suite of machine learning algorithms. We will use data from remote sensing and in-situ sources and study a wide range of predictors. The output would be a state-of-the-art system to estimate peak flowrates, which will be used in flood modeling. We will build the system in such a way that it can be updated easily in light of new data.
Objective Our specific objectives are (1) Identification and collection of new, high-resolution datasets relevant to the project from multiple sources; (2) Application of machine learning algorithms to identify relevant features from the selected datasets; (3) Application of machine learning algorithms to generate peak flowrates at different return periods and comparison of the outcomes with the existing schemes, i.e., the regression equations; (4) Addition of physically-based constraints to ensure realism; and (5) Evaluation of the viability of online training.
Impacts/Benefits Regional regression analysis is widely implemented in flood modeling. Regression equations used for Nebraska, however, are decades old. The three available sets of equations often produce results that vary by orders of magnitude. Therefore, there is a serious need to improve the accuracy of streamflow prediction methods using recent datasets and advanced methods. The outcome of this project is expected to significantly benefit flood forecasting, which in turn, will enhance transportation safety.
Deliverables
Download the Final Report