1 / 30

Predicting Hydrological Drought: Relative Contributions of Soil Moisture and Snow Information

Predicting Hydrological Drought: Relative Contributions of Soil Moisture and Snow Information To Seasonal Streamflow Forecast Skill. Randal Koster 1 , Sarith Mahanama 1 , Ben Livneh 2 , Dennis Lettenmaier 2 , and Rolf Reichle 1 1 Global Modeling and Assimilation Office, NASA/GSFC

salene
Download Presentation

Predicting Hydrological Drought: Relative Contributions of Soil Moisture and Snow Information

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Predicting Hydrological Drought: Relative Contributions of Soil Moisture and Snow Information To Seasonal Streamflow Forecast Skill Randal Koster1, Sarith Mahanama1, Ben Livneh2, Dennis Lettenmaier2, and Rolf Reichle1 1 Global Modeling and Assimilation Office, NASA/GSFC 2 Dept. of Civil and Env. Engineering, U. Washington Direct questions to: randal.d.koster@nasa.gov

  2. Streamflow Prediction Obvious: Larger snowpacka Increased streamflow during snowmelt season. Knowledge of winter snow a streamflow forecast skill

  3. Less obvious: Impact of soil moisture Snow (or rainfall) over wet soil: most of the meltwaterruns off into streams, reservoirs Snow (or rainfall) over dry soil: most of the meltwaterinfiltrates the soil and is lost to water resources Knowledge of winter soil moisture a streamflow forecast skill

  4. Quantify with experiment: 1. Perform multi-decadal offline simulation covering CONUS, using observations-based meteorological data. Determine streamflows in various basins and compare against (naturalized) streamflow observations. 2. Repeat, but doing forecasts: Simulate seasonal streamflow knowing only soil moisture and snow conditions on start date. (Use climatological met forcing for the forecast period.) Compare forecasts to observations. (Not a synthetic study!) 3. Repeat, knowing only snow conditions on the start date. 4. Repeat, knowing only soil moisture conditions on the start date.

  5. Four land models used, with results combined: • Catchment (GSFC) Vic (UW) • Noah (UW) Sac (UW) • Multiple start dates examined (Jan. 1, Feb. 1, …, Dec. 1) • Results are compared to naturalized streamflow observations for 23 basins in the U.S (between 39 and 83 years of data). Note: many of these basins are dominated by snowmelt runoff.

  6. Results (Exp1) for Snake River Basin: AMJ streamflow predicted from Apr. 1 initial conditions Red: forecast Blue: obs Z-score

  7. Results (Exp1) for Snake River Basin: AMJ streamflow predicted from Apr. 1 initial conditions Red: forecast Blue: obs Z-score Hydrological drought conditions forecasted! Henceforth, examine skill in predicting hydrological drought in the more general context of overall streamflow prediction skill (r2).

  8. Results for 3-month forecasts at zero lead Apr. 1 a AMJ Exp1: snow and soil moisture initialized Skill (r2)

  9. Results for 3-month forecasts at zero lead Apr. 1 a AMJ Exp1: snow and soil moisture initialized Exp2: snow initialized Exp3: soil moisture initialized Skill (r2)

  10. Skill as a function of start date: Results for 3-month forecasts at zero lead Jan. 1 a JFM Apr. 1 a AMJ Oct. 1 a OND July 1 a JAS Exp1: snow and soil moisture initialized Exp2: snow initialized Exp3: soil moisture initialized

  11. Skill as a function of start date: Results for 3-month forecasts at zero lead Jan. 1 a JFM Apr. 1 a AMJ Oct. 1 a OND July 1 a JAS Exp1: snow and soil moisture initialized Exp2: snow initialized Exp3: soil moisture initialized Outside of spring, soil moisture’s contribution to skill outweighs that of snow.

  12. Skill levels, however, must be considered in the context of seasonal flow volume and variability… Skill levels for Exp1 Fraction of annual flow

  13. Analysis… Streamflow during a forecast period is a function of: 1) Initial land surface moisture states (assume known) 2) Weather (precipitation) during the forecast period (assume unpredictable) Hypothesize: The relative magnitudes of the variances of initial land water content and forecast period precipitation have a first-order impact on achievable skill.

  14. (Extreme) Examples April 1st water storage: high variance April – June precipitation: low variance April – June streamflow is highly predictable April 1st water storage: low variance April – June precipitation: high variance April – June streamflow is not predictable

  15. Define a dimensionless diagnostic that characterizes the relative magnitudes of these variances: Standard deviation of total snow and soil water storage on day 1 of forecast Land models can now estimate this with reasonable accuracy! sW k = sPDt Standard deviation of total precipitation during forecast period Forecast skill should increase with k…

  16. “Synthetic truth” analysis. Using the streamflows produced in the control run (i.e., produced with known met. forcing) as “observations” allows an analysis across the contiguous United States. Quantification of k provides a picture of where in the U.S. skill may be achieved…

  17. Zero lead forecasts of 3-month streamflow (Exp1: Both snow and soil moisture initialized) skill level (r2) k JFM k distribution successfully describes, to first order, … …the places where skill (against synthetic truth) is achieved.

  18. Zero lead forecasts of 3-month streamflow (Exp1: Both snow and soil moisture initialized) skill level (r2) k JFM skill level (r2) k AMJ

  19. Zero lead forecasts of 3-month streamflow (Exp1: Both snow and soil moisture initialized) skill level (r2) k JAS skill level (r2) k OND

  20. Why NASA? One reason: soil moisture products will be made available through the SMAP mission. • For soil moisture, SMAP provides: • High revisit time (2-3 days) • High spatial resolution (10 km) • Depth to 5 cm (Level 2) • Depth through the root zone (Level 4, with data assimilation) % vol. soil moisture

  21. Summary 1. Today’s LSMs are capable of producing skillful seasonal forecasts (against real observations) of streamflow. 2. Snow information contributes more to skill during the snowmelt season. Soil moisture, however, contributes more skill (indeed, quite large levels) during the rest of the year. The results bode well for the usefulness of upcoming soil moisture datasets (e.g., from SMOS and SMAP). 3. Skill levels appear connected to k (i.e., sinitial-water/sprecipitation). Given the land models’ ability to estimate sinitial-water , this allows a quick estimation of where and when streamflow forecasting may be possible in areas outside the gauged basins.

  22. (Extra slides)

  23. Streamflow forecast skill (vs. obs) for Exp1: soil moisture and snow initialized Each dot corresponds to a (lead 0) 3-month forecast in one of the measured basins r2 k

  24. Streamflow forecast skill (vs. obs) for Exp1: soil moisture and snow initialized Red dots are the subset for which the control run (with known met. forcing during forecast period) does reasonably well vs. obs (r2 > 0.75). r2 k

  25. Preliminary k calculation from a single global run

  26. Missouri River at Hermann Snake River Initialize snow and soil water Total 3-month streamflow variance J F M A M J J A S O N D J F M A M J J A S O N D Initialize snow only Amount of variance explained J F M A M J J A S O N D J F M A M J J A S O N D Initialize soil water only J F M A M J J A S O N D J F M A M J J A S O N D

  27. Forecast skill (for 3-month totals) as a function of lead Arkansas River near Ralston Exp1: soil moisture and snow initialized

  28. Forecast skill (for 3-month totals) as a function of lead and start date Arkansas River near Ralston

  29. srunoff mrunoff k

More Related