Improving Data Anomaly Detection and Forecasting in the US Army Corps of Engineers Reservoir Sedimentation Information (RSI) Database

Description

I. Base Task 1: Increasing Composite Dataset for Analysis: For the previous analysis, only USACE and USBR reservoirs that had three or more surveys in the RSI system were included. This task will expanding the preliminary flagging, autonomous anomaly detection, and prediction model development to include all USACE and USBR reservoirs within the RSI system that have two or more surveys. In the current database, that includes 133 additional reservoirs. This number is expected to increase with the summer 2023 updates. Expanding the survey flagging process would specifically benefit data quality control for the RSI system. The additional flagging analysis will require quantifying reservoir pool elevations for capacity analysis and vertical datum correction factors for the 133+ additional USACE and USBR reservoirs.Note: USBR is in the process of developing their own independent database more suited to agency needs. At the time of this solicitation, there are plans to eventually connect the two databases for seamless updating. However, it is expected that contact and coordination with a USBR collaborator will be necessary to obtain the most recent data from that agency. II. Base Task 2: Demonstration of Forecasting Feasibility: Comparing model forecasted capacities with values from extrapolating known sedimentation rates will be a valuable tool for USACE to estimate economic, environmental, operational, and social impacts of future sedimentation and reservoirs. This comparison will serve as a validation to the proof of concept for predicting reservoir sedimentation using OLS and machine learning predictive models.An option may be awarded for FY25 based on sufficient progress and results of the base tasks in early FY24. III. Option Task 1: Developing Automated ArcGIS-based Analysis and Visualization Tools: Develop an ArcGIS Pro Add-In tool to predict capacity loss for a reservoir, from the date of dam construction to present day and use the results of Base Task 2 to predict future loss. The input data required for the capacity loss prediction models include the dam location, construction date, and original volume capacity. All other model parameters should be derived through automated GIS analysis. The tools should include a ‘water supply specific’ visualization tool component and use a stoplight ranking for reservoirs based on a number of fields from the database. This product should be developed to integrate with the RIMORPHIS geomorphic database while collaborating with USACE and other federal partners to identify a permanent, public-facing home for the database.

Agency Contact

More from Engineer Research and Development Center