Presentation is loading. Please wait.

Presentation is loading. Please wait.

Katrina Grantz, Balaji Rajagopalan, Edie Zagona, and Martyn Clark

Similar presentations


Presentation on theme: "Katrina Grantz, Balaji Rajagopalan, Edie Zagona, and Martyn Clark"— Presentation transcript:

1 Katrina Grantz, Balaji Rajagopalan, Edie Zagona, and Martyn Clark
INCORPORATING LARGE-SCALE CLIMATE VARIABILITY INFORMATION INTO WATER RESOURCES DECISION MAKING: A Case Study of the Truckee-Carson River System Katrina Grantz, Balaji Rajagopalan, Edie Zagona, and Martyn Clark OVERVIEW This poster summarizes recent work on using climate information to improve seasonal forecasts in the Truckee and Carson rivers. USBR needs good seasonal forecasts on Truckee and Carson Rivers (accurate and with long-lead time) to determine reservoir releases and inter-basin transfers through the Truckee Canal. CLIMATE DIAGNOSTICS Climate diagnostics are used to identify large-scale climate features related to flow in the Truckee and Carson rivers. Climatology Analysis Correlation Analysis- Winter Because the basins are snowmelt dominated we correlate spring runoff with climate signals over the Pacific Ocean the preceding winter and fall. Forecasting Results- Use of Climate Information in the Forecast Forecasting Results- Fall DECISION SUPPORT SYSTEM (DSS) Funding Provided By: CIRES USBR- Lahontan Basin Area Office Correlation Analysis- Fall Composite Analysis: High-Low Years STOCHASTIC FORECAST Forecasting Results- All Years Forecasts are improved when climate information (e.g. SST and geopotential height) is incorporated into the forecast. Forecasts without climate information are based on SWE alone. Wet and dry years are compared to examine the utility of the improved forecasts in years of extreme streamflow. Fall climate correlations exhibit a persistence in the atmospheric circulation patterns seen in the winter corrlation analysis. This enhances the prospect of longer lead forecasts. SWE alone SWE & Large-Scale Climate Figure 3. Carson River) spring streamfllow correlated with fall geopotential height 500mb (left) and sea surface temperature (right) Climate Diagnostics Truckee Canal Streamflow composites of the six wettest years minus the six driest years show the vector winds rotating counter-clockwise around the region off the coast of Washington (bringing moist air from the tropics to fall in the mountians as snow) and a SST pattern similar to that in the correlation plots. Figure 8. Forecasts without (left) and with (right) climate information in wet (top) and dry (bottom) streamflow years Diversions through the Truckee Canal must balance the interests of the Newlands Project farming district and the endangered cui-cui and Lahontan cutthroat trout Stocahstic Forecast Forecasts issued from fall are based on Pacific climate information alone. The forecasts demonstrate the ability to capture whether spring runoff will be above or below average six months in advance. Incorporation Into DSS Figure 4. Climate composites of sea surface temperature (left) and vector winds (right) for high minus low streamflow years in the Carson River Figure 9. Scatterplots of median of fall ensemble forecast vs. observed value in the Trucee (left) and Carson (right) rivers. We utilize the nonparametric techniques of the modified k-nearest neighbor (K-NN) algorithm to generate ensemble forecasts of spring streamflow. Predictors, based on climate diagnostics, are SWE, SST, and 500mb geopotential height from regions of highest correlation. To test the utility of the new forecasting technique and impacts on operations, forecasts will be input to the Truckee Operations RiverWare model used by the USBR. The annual cycles of streamflow and precipitation indicate that the basins are primarily snowmelt dominated. The modified K-NN method uses a local polynomial fit for the mean forecast and bootstraps the residuals to generate an ensemble forecast. assumes no underlying structure to the data (no transformations needed) produces flows not seen in the historical record perturbs the historical record within its representative neighborhood Yt=yt*+et* Figure 1. Average monthly flow in the Truckee and Carson rivers (left) and average monthly precipitation in the Sierra Nevada climate division. yt* et* Water Available For Irrigation Spring streamflow in the Truckee and Carson rivers correlates strongly with the 500mb geopotential height variable in the region off the coast of Washington and sea surface temperature in the northern Pacific the preceding winter. Regions of high correlation are the same for the Truckee and Carson rivers. Figure 5. Figure illustrating local polynomial fit and residual resampling for modified K-NN method Diversion Through Truckee Canal The forecasting method is validated using standard cross-validation: dropping one point from the forecasting model and then predicting it. Figure 10. Truckee RiverWare model (above) and seasonal forecasting model results (right) for a dry year The solid line represents model results using the ensemble forecast, the dashed line represents model results using the climatological forecast. The point depicts the value that would have occurred given a perfect forecast. Water Available For Fish Seasonal Operations Model Because the Truckee RiverWare model is not fully operational, we develop a simplified seasonal operations model. We test decision variables such as Water Available for Irrigation, Diversion through the Truckee Canal, and Water Available for Fish. R=0.87 R=0.92 Figure 6. Forecasting results for Truckee (top) and Carson (bottom) rivers for entire period of record. The solid line represents the observed value, the boxplots represent the ensemble forecast in each year and the horizontal lines represent the 5th, 25th, 50th, 75th, and 95th percentile of the historical data. Figure 7. Scatterplots of median of ensemble forecast vs. observed value for each year in the Truckee River (left) and Carson River (right). Figure 2. Carson River (top) and Truckee River (bottom) spring streamflow correlated with winter geopotential height 500mb (left) and sea surface temperature (right)


Download ppt "Katrina Grantz, Balaji Rajagopalan, Edie Zagona, and Martyn Clark"

Similar presentations


Ads by Google