Award Date

12-2010

Degree Type

Dissertation

Degree Name

Doctor of Philosophy in Engineering

Department

Civil and Environmental Engineering

First Committee Member

Thomas C. Piechota, Chair

Second Committee Member

Jacimaria Batista

Third Committee Member

Sajjad Ahmad

Fourth Committee Member

Ashok K. Singh

Graduate Faculty Representative

Susanna Priest

Number of Pages

265

Abstract

The National Weather Service’s (NWS) river forecast centers provide long-term water resource forecasts for the main river basins in the U.S. The NWS creates seasonal streamflow forecasts using an ensemble prediction model called the Extended Streamflow Prediction (ESP) software. ESP creates runoff volume forecasts by taking the current observed soil moisture and snowpack conditions in the basin and applying them to historical temperature and precipitation scenarios. The ESP treats every historic input year as a likely scenario of future basin conditions. Therefore improving the knowledge about how long-term climate cycles impact streamflow can extend the forecast lead time and improve the quality of long-lead forecasts.


First, a study of the existing climate indices is carried out in Chapter 3 to establish which index shows a significant long-lead connection to the Colorado River Basin (CRB). Using Singular Value Decomposition (SVD) this step identifies a 1-year lagged relationship between the Pacific Ocean sea surface temperatures (SST) and CRB streamflow. A new SST region is identified in this analysis (named the Hondo region) and compared to the other established climate indices (e.g. SOI, PDO, NAO, AMO). The tests demonstrate Hondo performs better at longer lead times than the existing climate indices.

Second, Chapter 4 identifies the climate cycles impacting the CRB streamflow. The SVD analysis performed in Chapter 3 is extended to include the simultaneous (or 0-year lag) as well as the 2nd and 3rd year lag times. Because SST’s and streamflow are basically independent data, this chapter explains the physical connection between them by analyzing the relationship between the ocean, atmosphere and CRB streamflow. This analysis, as well as recent research into the Pacific Quasi-Decadal Oscillation (PQDO), demonstrates that the CRB streamflow is dominated by a hierarchy of climate drivers. The Hondo is the secondary level which exists between the extreme impacts of the ENSO signal and the longer cyclical patters observed in the PQDO. The current research shows that the QDO cycle leads precipitation by three years, and the Hondo leads the CRB by one to two years.

Finally, the information from Chapter 4 reveals that the Hondo region can be used as a basis for weighing the ESP output. This is done because water resource managers create multi-year water plans that are utilized to project power generation supplies and water system improvements. In Chapter 5 several methods of weighing the ESP output based on the Hondo region are presented. Each method is assessed using parametric and non-parametric forecast skill metrics. The overall goal of this chapter is to identify a weighting technique, lag time, and season interval which show a marked improvement over simply using the 30-year mean as a streamflow predictor. From this analysis the forecast skill score is optimized when using the January – March average SST values as a basis for forecasting the streamflow for the following water year.

Keywords

Climate; Climatology; Forecast; North America – Colorado River; Runoff – Forecasting; Singular value decomposition; Streamflow – Forecasting; Water supply – Forecasting decomposition; Water supply

Disciplines

Civil and Environmental Engineering | Climate | Environmental Monitoring | Environmental Sciences | Water Resource Management

Language

English