Presentation is loading. Please wait.

Presentation is loading. Please wait.

Advances in hydrologic prediction in the western U.S. Dennis P. Lettenmaier Department of Civil and Environmental Engineering University of Washington.

Similar presentations


Presentation on theme: "Advances in hydrologic prediction in the western U.S. Dennis P. Lettenmaier Department of Civil and Environmental Engineering University of Washington."— Presentation transcript:

1 Advances in hydrologic prediction in the western U.S. Dennis P. Lettenmaier Department of Civil and Environmental Engineering University of Washington Catchment-scale hydrological modeling and data assimilation international workshop (CAHMDA) III Melbourne, Australia January 10, 2008

2 Talk Outline 1)Background 2)Seasonal hydrologic prediction approaches 3)The University of Washington west-wide seasonal hydrologic forecast system 4)Current and recent research a)Assimilation of satellite data b)Is calibration necessary? c)Processing of real-time forcings d)Is there hydrologically relevant skill in climate forecasts? 5)Conclusions and challenges

3 1. Background: The importance of Seasonal Hydrologic Forecasting water management hydropower irrigation flood control water supply fisheries recreation navigation water quality AugDecApr Reservoir Storage Aug

4 2. Seasonal hydrologic prediction approaches Statistical and stochastic methods –Simple to apply –Accuracy depends on sample size –Difficulties with extremes and nonstationarity Dynamic hydrological modeling (via Ensemble Streamflow Prediction) – most of this talk –Requires model calibration (Achilles Heel of hydrologic prediction –More or less immune from deficiencies of statistical and stochastic methods –Consistency with coupled modeling approaches used in numerical weather and climate prediction Hybrid approaches –Not widely used, but deserving of more attention

5 Snow water content on April 1 April to August runoff McLean, D.A., 1948 Western Snow Conf. SNOTEL Network Application of statistical methods to seasonal hydrologic prediction in the western U.S. PNW

6 Typical SNOTEL Site

7 Overview: ESP Hydrologic prediction strategy ESP data flow The ESP “spider web”

8 ESP Implementation by NWS – Lake Tahoe inflow forecasts

9 3. The University of Washington west-wide seasonal hydrologic forecast system 6-month ESP streamflow forecasts for western U.S. and Mexico effective 1/1/08

10

11 UW Seasonal Hydrologic Forecast System Website

12 Forecast System Initial State information Soil Moisture Simulated Initial Condition Snowpack Simulated Initial Condition Observed SWE

13 Assimilation Method weight station OBS’ influence over VIC cell based on distance and elevation difference number of stations influencing a given cell depends on specified influence distances spatial weighting function elevation weighting function SNOTEL/ASP VIC cell Forecast System Initial State Snow Adjustment distances “fit”: OBS weighting increased throughout season OBS anomalies applied to VIC long term means, combined with VIC- simulated SWE adjustment specific to each VIC snow band

14 Streamflow Forecast Details Flow location maps give access to monthly hydrograph plots, and also to raw forecast data. Clicking the stream flow forecast map also accesses current basin- averaged conditions

15 Streamflow Forecast Results: Westwide at a Glance

16 Winter 2006-07: seasonal volume forecast for APR-SEP OBS Forecasts made on 1 st of Month

17 “User” Interactions associated with these research applications UW Hydrologic Forecast and Nowcast Systems U. Arizona / USBR forecast study, Lower Colorado basin NWS Hydrologic Ensemble Prediction Experiment 3TIER Environmental Forecast Group NRCS National Water and Climate Center NOAA Climate Prediction Center’s US Drought Outlook Miscellaneous: Seattle City Light, energy traders, hydropower utilities, NOAA regional climate offices UW Rick Palmer Group Puget Sound region flow forecasts UW Climate Impacts Group (CIG) Annual Water Outlook meetings NOAA National Centers for Environmental Prediction (NCEP) testbed activities Columbia River Inter-tribal Fish Commission Klamath R. Basin Bureau of Reclamation UCI / California Dept of Water Resources WA State Dept of Ecology & Yakima R. Basin Bureau of Reclamation new US Drought Monitor Princeton University Hydrologic Forecast System

18 4a: Current and recent research: Snow data assimilation

19

20

21 NCDC met. station obs. up to 2-4 months from current local scale weather inputs Initial Conditions: soil moisture, snowpack Hydrologic model spin up MODIS Update Ensemble Forecast: streamflow, soil moisture, snowpack, runoff 25 th Day of Month 01-2 years back LDAS/other real-time met. forcings for remaining spin-up Hydrologic simulation End of Month 6 - 12 MODIS updating of snow covered area Snowcover before MODIS updateSnowcover after MODIS update Change in Snowcover as a Result of MODIS Update for April 1, 2004 Forecast

22 Unadjusted vs adjusted forecast errors, 2001- 2003, for reservoir inflow volumes (left plot) and reservoir storage (right)

23 Passive microwave remote sensing for snow water equivalent In principle, attractive since it measures the “right” variable (water equivalent rather than extent) AMSR-E product probably is best current generation, but numerous problems (mostly generic): –Coarse resolution (~ 15-25 km) –Saturation at 100-200 mm SWE –Requires dry snowpacks (algorithms fail if there is liquid water in the pack) –Algorithms unreliable for mixed pixels (especially forest) –Signature is highly sensitive to grain size (and other snow microphysical properties)

24 4b: Current and recent research: Is calibration necessary? Approach: Use percentile mapping bias correction on uncalibrated forecasts, compare Cp (1 – forecast MSE/unconditional variance) for calibrated and uncalibrated (using N-LDAS parameters) for a range of forecast dates and lead times at 8 forecast sites throughout the western U.S. Result: Bias corrected uncalibrated forecasts did nearly as well at most sites, and better at some Conclusion: Perhaps calibration (the Achilles Heel of dynamic hydrologic forecasting methods) isn’t really necessary

25

26

27 4c: Processing of real-time forcings Problem: Generally there are many fewer real-time stations (station numbers shown in figure) than retrospective (all stations

28 Question: How to interpolate data from real-time stations so as to respect orographic variations in precipitation (most stations are at low elevation, most precipitation at high elevation)? Options tested: a) Grid real-time stations directly, with PRISM orographic adjustment b) interpolate percentiles to locations of non-reporting stations, then grid data as if all had reported

29 Generally, percentile interpolation methods (“Index”) do better than direct interpolation methods (“Maurer”)

30 4d: Is there hydrologically relevant skill in climate forcings

31 Results of previous seasonal hydrologic predictability studies for continental U.S. 1.Wood et al, “A retrospective assessment of NCEP climate model-based ensemble hydrologic forecasting in the western United States”, JGR, 2005 2.Wood, “An ensemble-based framework for characterizing sources of uncertainty in hydrologic prediction” 3.Work in progress, DEMETER forecasts over continental U.S.

32 Wood et al 2005: Retrospective Assessment: Results using GSM General finding is that NCEP GSM climate forecasts do not add to skill of ESP forecasts, except… April GSM forecast with respect to climatology (left) and to ESP (right)

33 Wood et al 2005: Retrospective results for ENSO years October GSM forecast w.r.t ESP: unconditional (left) and strong-ENSO (right) Summary: During strong ENSO events, for some river basins (California, Pacific Northwest) runoff forecasts improved with strong-ENSO composite; but Colorado River, upper Rio Grande River basin RO forecasts worsened.

34 Wood (2002) Reverse ESP

35 Reverse ESP vs ESP – typical results for the western U.S. Columbia R. Basin Rio Grande R. Basin ICs more impt fcst more impt

36 DEMETER forecast evaluation VIC model long-term (1960-99) simulations at ½ degree spatial resolution assumed to be truth DEMETER reforecasts with ECMWF seasonal forecast model for 6 month lead, forecasts made on Feb 1, May 1, Aug 1, Nov 1 1960-99 9 forecast ensembles on each date Forecast forcings (precipitation and temperature) downscaled and bias corrected using Wood et al approach (also incorporated in UW West-wide system) On each forecast date, 9 ensemble members also resampled at random from 1960-99 to form ESP ensemble Forecast skill evaluated using Cp for unrouted runoff

37 Test sites

38 Missouri River at Fort Benton

39 Snake River at Milne

40 Owyhee River

41 Riley

42 Rosco

43 Luo, L. and E. F. Wood (2007): Seasonal Hydrologic Prediction with the VIC Hydrologic Model for the Eastern U.S. Journal of Hydrometeorology. In review. USGS streamflow gauges that are used in the evaluation of streamflow predictions

44 The evaluation of streamflow predictions over selected gauges. The ranked probability score (RPS) for monthly streamflow for the first three months are examined against the offline simulation. The bars are for CFS, CFS+DEMETER and ESP from the left to the right, respectively. RPS: 0~1 with 0 being the perfect forecast 3 tercels, below normal, normal and above normal with probability of 1/3 each. Luo, L. and E. F. Wood (2007): Seasonal Hydrologic Prediction with the VIC Hydrologic Model for the Eastern U.S. Journal of Hydrometeorology. In review.

45 Concluding thoughts Hydrologic prediction skill at S/I lead times comes mostly from initial conditions. Hence more focus on data assimilation, and its implications for hydrologic forecast skill, needs more attention. The role of model error in hydrologic predictions needs more focus – how do we best weight land models in multimodel ensemble? Do hydrologists (and the land data assimilation community) need to expend more effort on hydrologic forecasting?


Download ppt "Advances in hydrologic prediction in the western U.S. Dennis P. Lettenmaier Department of Civil and Environmental Engineering University of Washington."

Similar presentations


Ads by Google