1 Water Supply Forecasting Research at UNLV Thomas Piechota, PhD, P.E. Director of Sustainability and Multidisciplinary Research Associate Professor, Department of Civil and Environmental Engineering Kenneth Lamb, P.E. Funded By: NOAA
Outline Climate Variability and the Colorado River Basin Existing Research Basis Current Research Goal Establishing Long Lead Climate/Streamflow teleconnections Develop Weighted Forecasting Method 2
Climate Variability Linkages of Colorado River basin hydrology with: Interannual climate variability (2-7 years) El Niño - Southern Oscillation (ENSO) Northern Atlantic Oscillation (NAO) Pacific Ocean Sea Surface Temperatures (SSTs) Interdecadal climate variability (10-20 years) Atlantic Multidecadal Oscillation (AMO) Pacific Decadal Oscillation (PDO) Motivation: Improved long range(6-24 months in advance) streamflow forecasting 3
Climate Indices - Streamflow Timilsena, Piechota, Tootle, & Singh, 2009
Pacific Ocean SST - Streamflow Tootle & Piechota, 2005
Pacific Ocean SST - Snowpack Aziz, Tootle, Piechota, Gray, 2009 (+) (-) (+)
Research Project Goal Improve the “Out-Year” inputs used by USBR Identify long-lead climate-streamflow connections Develop weighted forecast approach 7 ESP OUTLOOK ESP 3-MONTH APR-JUL COORDINATED USBR INTERPOLATION AVERAGE
Research Project Goal Improve the “Out-Year” inputs used by USBR Identify long-lead climate-streamflow connections Develop weighted forecast approach 8
Analysis Lagged Rank Sum Testing SOI, PDO, NAO, AMO Individually and ENSO-Coupled Lagged Singular Value Decompostion (SVD) Pacific Ocean SSTs Generate weighted forecasts and compare with: 30-Year Average ESP Outlook 9
10 Identify the Strongest Lag
Rank Sum Test Results 11
12 Developing SST Teleconnection Singular Value Decompostition (SVD) - Identifies covariance between two datasets Pacific Ocean Sea Surface Temperature Data (1951 – 2003) CRB Naturalized/Unimpaired Streamflow Data (1952 – 2004)
13 3-Month SST Avg SVD Results
Why This Region? Yang et al. 2002: Change in SST is evidence of a strong/weak wintertime jet stream Why the 1-year lag? Anybody want a Masters degree? 14
15 Forecasting With SST Index Forecast created using a weighted resampling of the observed data ( ) Four Different Weighting Formulas 3-Month Avg. SST intervals: DJF, JFM, FMA Predictands: April-July Volume, Yearly Volume
16 Assessing Forecast Skill Linear Error in Probability Space (LEPS) score Compares forecast to the observed mean Probability of Observed values compared with probability of Forecasted values: QfQf QOQO PfPf POPO
17 Skill Results (DJF | Lag 1 year) 30-yr average skill scores at each location (PRELIMINARY)
Conclusions Improvement to the 30-Year Avg using Hondo Remaining Work: Quantify Improvement to the ESP Outlook 18