Overview of Hydrologic Forecasting Research at the University of Washington Alan F. Hamlet, Andy Wood, Dennis P. Lettenmaier JISAO/CSES Climate Impacts Group Dept. of Civil and Environmental Engineering University of Washington
Overview of the UW West-Wide Hydrologic Forecasting System
Fundamental Research Goals: Implement a consistent set of fully distributed hydrologic simulation models and forecasting approaches over the entire West Improve estimates of hydrologic initial conditions (SWE and soil moisture) via improved use of real time station records and data assimilation using remotely sensed (MODIS snow extent) and telemetered (SNOTEL SWE) snow observing systems. Make use of improved climate information and forecasts (e.g. ENSO forecasts, climate model simulations) at number of time scales using a number of approaches
Experimental W. US Hydrologic Forecast System NCDC met. station obs. up to 2-4 months from current local scale (1/8 degree) weather inputs soil moisture snowpack Hydrologic model spin up SNOTEL Update streamflow, soil moisture, snow water equivalent, runoff 25th Day, Month 0 1-2 years back LDAS/other real-time met. forcings for spin-up gap Hydrologic forecast simulation Month 6 - 12 INITIAL STATE SNOTEL / MODIS* Update ensemble forecasts ESP traces (40) CPC-based outlook (30) NCEP CFS ensemble (N) NSIPP/GMAO ensemble (9) This just describes the framework for simulations – how the forecasts are made. Note that the update cycle is typically monthly, with forecasts initialized on day 1 of the month, although we occasionally do twice monthly forecasts if changing conditions warrant. * experimental, not yet in real-time product
Experimental W. US Hydrologic Forecast System Snowpack Initial Condition Here I typically say the system is implemented using VIC (wat/nrg balance model, etc.), and the spinup simulation produces an initial condition for snowpack and soil moisture over a domain of about 18,000 cells. Soil Moisture Initial Condition
Experimental W. US Hydrologic Forecast System Multiple Seasonal Climate Forecast Data Sources CCA NOAA CAS OCN CPC Official Outlooks SMLR … Coupled Forecast System VIC Hydrology Model NASA NSIPP/GMAO dynamical model Our framework has allowed the use of climate forecast information from a number of sources – ESP & derivatives, climate model (NCEP/NSIPP), and CPC probabilistic outlooks. Currently we are re-implementing the climate model forecasts, and the CPC forecast approach is also evolving as we try to improve the temporal disaggregation step. ESP ENSO UW ENSO/PDO
A Walk Through the UW West Wide Hydrologic Forecasting Web Site: http://www.hydro.washington.edu/forecast/westwide/spatial/index.shtml
Some Related UW Forecasting Research Projects: Daily Surface Water Monitor: Daily updated SWE and SM product for the U.S. [www.hydro.washington.edu/forecast/monitor/index.shtml] Multi-Model Ensemble Forecasting Methods: Sacramento Model VIC NOAH Increased VIC Spatial Resolution Over a Smaller Domain 1/16th degree for WA, OR, ID, NV Yakima basin Klamath basin
Interaction with NRCS NWCC Since last year, we have exchanged nowcast/forecast results with the NRCS National Water and Climate Center (head: Phil Pasteris) Under a Memorandum of Understanding between NRCS & UW: UW provides forecast results and data as NRCS requests NRCS provides access to stream flow and climate data (primarily via NOAA ACIS) NRCS has created a place for links to “experimental water supply forecasts” from its official website. Currently the UW is the only one, and they would like more! We generally attempt to schedule a “pre-forecast” conference call just prior to NRCS coordination of forecasts with NWS RFCs, in which we summarize our forecast outlooks and compare notes. In addition, there is a fair amount of informal exchange. We also have an on-going interaction with the NWCC, and exchange results and comments on a routine basis during the forecast season. Credit Phil Pasteris, Tom Pagano, Tom Perkins, Jolyne Lea. Note that NWCC is experimenting with a modeling capability based on the USGS PRMS model.
Overview of UW Climate Research
Pacific Decadal Oscillation El Niño Southern Oscillation A history of the PDO A history of ENSO warm warm cool 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
Effects of the PDO and ENSO on Columbia River Summer Streamflows PDO Cool Cool Warm Warm Red=warm ENSO Green=ENSO neutral Blue=cool ENSO
1998 ✔ 1999 ✔ 2000 ✔ 2001 X 2002 ✔ 2003 ✔ 2004 X 2005 ✔ In 6 out of 8 test years, accurate categorical ENSO forecasts (warm, neutral, cool) have been available in mid-summer preceding the water year. By October simple persistence gives an accurate forecast.
Evaluation of Real Time Nino3.4 Forecast for WY 2005 Observed Nino3.4 anomally
Natural Streamflow (cfs) Bias Corrected Long Range Streamflow Forecast for the Columbia River at The Dalles January Nino3.4 index between 0.2 and 1.2 Red = Unconditional mean Blue = Ensemble mean Black = 2005 Observed Natural Streamflow (cfs)
Comparison of Skill For Warm ENSO Years
Comparison of Skill For Cool ENSO Years
Comparison of Skill For ENSO Neutral Years
X100 wENSO / X100 2003 X100 nENSO / X100 2003 X100 cENSO / X100 2003 Fig 8 Warm, neutral, cool ENSO 100-year flood DJF Avg Temp (C) DJF Avg Temp (C) DJF Avg Temp (C) X100 wENSO / X100 2003 X100 nENSO / X100 2003 X100 cENSO / X100 2003
Some Current Users of Long-Range Climate Forecasts for Hydrologic Forecasting Seattle Public Utilities Portland General Electric 3-Tier Environmental Forecast Group Seattle City Light Tacoma Power Columbia Basin USACE Libby and Dworshak forecasts (SOI)
At almost every USHCN station, winters warmed + signs: warming but not statistically significant
TMAX Regionally Averaged Cool Season Temperature Anomalies 0.74 0.63 0.76 0.62 (Regional to Global Correlation R2 ) TMAX
TMIN Regionally Averaged Cool Season Temperature Anomalies 0.84 0.87 0.94 0.73 (Regional to Global Correlation R2 ) TMIN
Simulated Changes in Snowpack Timing in the Western U.S. a) 10 % Accum. b) Max Accum. c) 90 % Melt 1916-2003 Effects of Temperature and Precipitation DJF Temp (C) DJF Temp (C) DJF Temp (C) Change in Date Change in Date Change in Date Effects of Temperature only DJF Temp (C) DJF Temp (C) DJF Temp (C) Change in Date Change in Date Change in Date Effects of Precipitation only DJF Temp (C) DJF Temp (C) DJF Temp (C) Change in Date Change in Date Change in Date
Flood Control vs. Refill Streamflow timing shifts can reduce the reliability of reservoir refill Full Model experiments (see Payne et al. 2004) have shown that moving spring flood evacuation two weeks to one month earlier in the year helps mitigate reductions in refill reliability associated with streamflow timing shifts. Payne, J.T., A.W. Wood, A.F. Hamlet, R.N. Palmer, and D.P. Lettenmaier, 2004, Mitigating the effects of climate change on the water resources of the Columbia River basin, Climatic Change, Vol. 62, Issue 1-3, 233-256
Simulated Changes in the 20-year Flood Associated with 20th Century Warming DJF Avg Temp (C) X20 2003 / X20 1915 Fig 3 20 year flood A spatial scale DJF Avg Temp (C) X20 2003 / X20 1915 X20 2003 / X20 1915
Regionally Averaged Cool Season Precipitation Anomalies
20-year Flood for “1973-2003” Compared to “1916-2003” for a Constant Late 20th Century Temperature Regime DJF Avg Temp (C) X20 ’73-’03 / X20 ’16-’03 X20 ’73-’03 / X20 ’16-’03
Summary: 1) Overview of UW West-Wide Hydrologic Forecasting System: Some overall research goals of the project are to implement a set of consistent hydrologic simulation models and forecasting procedures over the West, improve spatially explicit estimates of initial hydrologic conditions by accessing real time station and assimilating measured SWE data, and to explore the use of climate information and forecasts using a number of approaches. Some related current research projects include implementation of multi-model ensembles to improve forecasting accuracy, daily updating of SM and SWE fields for the continental US for the US Drought Monitor, and a four-state higher spatial resolution forecasting system with pilot applications in the Yakima and Klamath basins.
Summary: 2) Use of Climate Information and Forecasts: Long range ENSO and PDO forecasts have been shown in retrospective tests to improve ESP forecast skill and to extend the practical lead time of long-range streamflow forecasts at The Dalles. Some operational statistical forecasting procedures currently include ENSO information (via SOI). ENSO also appears to affect flood risks in the PNW, particularly in coastal areas. Temperature and precipitation variability have been changing over the 20th century. The late 20th century is warmer and has different cool season precipitation variability than the pre-1975 period. Rising temperatures have resulted in reduced and earlier snowmelt (e.g. affecting refill timing). Both effects have been shown in simulations to result in altered flood risks over the West. Are current flood control practices robust to these changes?