Identification of Pacific Ocean Sea Surface Temperature Teleconnections with Western United States Streamflow

G. Tootle
T. C. Piechota, University of Nevada Las Vegas

Presented at the American Geophysical Union (AGU) Fall Meeting, December 8-12, 2003, San Francisco, California.


The western United States is currently experiencing a significant drought which is causing water shortages and low lake levels. This analysis will provide information on Pacific Ocean Sea Surface Temperatures (SSTs) as a predictor variable of Western United States streamflow. The Partial Least Squares (PLS) statistical technique will be utilized to determine the best predictor ranges. PLS, which is primarily used in chemical spectrometry analysis, is an extension of multiple linear regression. PLS is an exploratory tool used to select suitable predictor variables and to identify outliers. Streamflow data (water year total runoff volume in acre-feet) were obtained from the U.S. Geological Survey (USGS) NWISWeb Data retrieval ( for sixty four unimpaired streamflow stations in the Western United States from 1946 to 2001 (56 years of data). Principal Component Analysis was performed on the streamflow data to determine areas in which the streamflow stations behaved similarly. The time series for these components were then utilized as the predictand. SST data were obtained from the National Climatic Data Center website ( The SST data consists of average monthly values for a 2 degree by 2 degree grid cell and the range of data used for the analysis was Latitude 120 degrees West to Latitude 80 degrees East and Longitude 70 degrees South to Longitude 70 degrees North. This resulted in a grid with 81 cells in the x-direction and 71 cells in the y-direction (5,751 cells). The SST predictors cover a period from 1945 to 2000. The average monthly values of the SST predictors are averaged for each season: April-May-June (AMJ - spring season), July-August-September (JAS - summer season) and October-November-December (OND - fall season). The best long lead time (3 to 9 months) indicators for water supply forecasting will be identified.