1. Application of NLDAS Ensemble LSM Simulations to Continental-Scale Drought Monitoring Brian Cosgrove and Charles Alonge SAIC / NASA GSFC Collaborators: Kenneth Mitchell 1, Kingtse Mo 2, Eric Wood 3 1 NOAA/NCEP/EMC, 2 NOAA/NCEP/CPC, 3 Princeton University Funded by NOAA CPPA and NASA WMP

2. Introduction Accurate drought characterization is vital to drought impact assessment and amelioration Accurate drought characterization is vital to drought impact assessment and amelioration Wide range of drought indices currently exist, each with its own strengths and weaknesses Wide range of drought indices currently exist, each with its own strengths and weaknesses Difficult to calibrate and improve upon certain indices due to a lack of long term and spatially continuous soil moisture observations on large scale Difficult to calibrate and improve upon certain indices due to a lack of long term and spatially continuous soil moisture observations on large scale Land Data Assimilation Systems (LDAS) offer high quality soil moisture fields with good spatial and vertical resolution and are a potentially useful tool in monitoring droughts Land Data Assimilation Systems (LDAS) offer high quality soil moisture fields with good spatial and vertical resolution and are a potentially useful tool in monitoring droughts Combine modeling infrastructure of North American LDAS (NLDAS) with long term (28 years+) forcing fields of NOAA’s North American Regional Reanalysis (NARR) to form a NARR-based NLDAS drought monitor Combine modeling infrastructure of North American LDAS (NLDAS) with long term (28 years+) forcing fields of NOAA’s North American Regional Reanalysis (NARR) to form a NARR-based NLDAS drought monitor

3. NASA GSFC Drought Project Outline Construct and validate 1/8th degree forcing dataset based on NARR, supplemented with observed precipitation, and bias corrected with observed radiation Construct and validate 1/8th degree forcing dataset based on NARR, supplemented with observed precipitation, and bias corrected with observed radiation Execute and validate 1/8th degree 28 year-long ensemble runs using Noah, CLM3, Mosaic, HySSiB, and Catchment LSMs Execute and validate 1/8th degree 28 year-long ensemble runs using Noah, CLM3, Mosaic, HySSiB, and Catchment LSMs Construct and execute drought monitor processing system using individual as well as 7 member ensemble output (includes VIC and Sacramento output from NCEP) Construct and execute drought monitor processing system using individual as well as 7 member ensemble output (includes VIC and Sacramento output from NCEP) Analyze drought monitor output to determine effect of model selection, forcing data, NARR climatology length, and ensemble construction on drought characterization, and to determine performance versus existing drought monitoring systems Analyze drought monitor output to determine effect of model selection, forcing data, NARR climatology length, and ensemble construction on drought characterization, and to determine performance versus existing drought monitoring systems Transition system to real-time operations, providing objective data to existing drought monitoring efforts such as the U.S. Drought Monitor where possible Transition system to real-time operations, providing objective data to existing drought monitoring efforts such as the U.S. Drought Monitor where possible Forcing Data LSM Runs Drought Monitor

4. Project Overview

5. Forcing Data and Ensemble Simulations Forcing compatible with current 1/8 th degree NLDAS systems Forcing compatible with current 1/8 th degree NLDAS systems NARR model data base (3 hourly, 32km, 1979 – Present) NARR model data base (3 hourly, 32km, 1979 – Present) Hourly SW bias correction developed from GOES data for each month and applied to NARR SW fields Hourly SW bias correction developed from GOES data for each month and applied to NARR SW fields Hourly observed precipitation based on daily PRISM gauge data, and hourly Stage II Doppler Radar, CMORPH, and HPD data. Hourly observed precipitation based on daily PRISM gauge data, and hourly Stage II Doppler Radar, CMORPH, and HPD data. Ensemble simulations with seven LSMs Ensemble simulations with seven LSMs Multi-model output will form base of drought monitor, and aid in LSM improvement as current NLDAS runs have done Multi-model output will form base of drought monitor, and aid in LSM improvement as current NLDAS runs have done Noah, CLM3, HySSiB, Catchment and Mosaic LSMs (plus VIC and Sacramento simulations from NCEP) Noah, CLM3, HySSiB, Catchment and Mosaic LSMs (plus VIC and Sacramento simulations from NCEP) 28 Years (1979-Present), 3-hourly on 1/8 th degree grid 28 Years (1979-Present), 3-hourly on 1/8 th degree grid Runoff routing scheme applied to each LSM’s output to calculate stream flow (Lohmann 1998) Runoff routing scheme applied to each LSM’s output to calculate stream flow (Lohmann 1998) Ensemble mean and individual LSM output intercompared and validated against observations and CPC 50 year Noah simulation Ensemble mean and individual LSM output intercompared and validated against observations and CPC 50 year Noah simulation

6. Wtd/UnWtd PDSI LSM Percentile Drought Monitor Processing System Drought monitor will compute several drought indices from NLDAS LSM output, NARR land surface states, and forcing Drought monitor will compute several drought indices from NLDAS LSM output, NARR land surface states, and forcing Standard and new NLDAS-based drought indices will be computed to assess agricultural, meteorological, and hydrological droughts, which only sometimes overlap Standard and new NLDAS-based drought indices will be computed to assess agricultural, meteorological, and hydrological droughts, which only sometimes overlap Self Calibrating (duration and climate characteristic parameters) Selection of indices is a key area for drought community input Selection of indices is a key area for drought community input Which indices are most useful, and most accurate? Which indices are most useful, and most accurate?

7. Analyze Drought Monitor Index Output How does the characterization of drought vary by LSM? How does the characterization of drought vary by LSM? What impact does use of the ensemble mean versus single model output have on drought detection? What impact does use of the ensemble mean versus single model output have on drought detection? How do drought indices produced by the ensemble LSMs compare to drought index values produced directly from NARR land surface fields? How do drought indices produced by the ensemble LSMs compare to drought index values produced directly from NARR land surface fields? Can a NARR/NLDAS system produce standard and experimental-LDAS drought index fields which capture the same droughts detected by established measures such as PSDI and US Drought Monitor? Can a NARR/NLDAS system produce standard and experimental-LDAS drought index fields which capture the same droughts detected by established measures such as PSDI and US Drought Monitor? How does climatology-length affect drought characterization, and does the NARR offer a suitably accurate and lengthy record of forcing data to serve as the base of a drought monitor? How does climatology-length affect drought characterization, and does the NARR offer a suitably accurate and lengthy record of forcing data to serve as the base of a drought monitor? New NOAA Climate Test Bed proposal with NCEP partners submitted to further address issue New NOAA Climate Test Bed proposal with NCEP partners submitted to further address issue Will also address need for drought forecasts Will also address need for drought forecasts

8. Current Project Status Second year of three year project Second year of three year project Ongoing collaborations with US Drought Monitor, NOAA NCEP, and Princeton University Ongoing collaborations with US Drought Monitor, NOAA NCEP, and Princeton University Forcing data generation complete Forcing data generation complete Real-time beta drought monitor constructed on NLDAS website Real-time beta drought monitor constructed on NLDAS website Mosaic and Noah test runs performed, highlighting several key issues for further investigation Mosaic and Noah test runs performed, highlighting several key issues for further investigation Climatology length Climatology length Meteorological forcing data Meteorological forcing data Model selection Model selection

9. NLDAS Experimental Drought Monitor Methodology Mean root zone and total column soil wetness values computed for each day of the year from the 1997-2007 NLDAS Mosaic and Noah output Mean root zone and total column soil wetness values computed for each day of the year from the 1997-2007 NLDAS Mosaic and Noah output Anomalies are computed by comparing the near real-time data to the same section of the year in the mean climatology files Anomalies are computed by comparing the near real-time data to the same section of the year in the mean climatology files Percentiles are computed by ranking the current soil wetness values against values from +/- 5 surrounding days over the past 10 years. Percentiles are computed by ranking the current soil wetness values against values from +/- 5 surrounding days over the past 10 years. Follows in footsteps of existing websites (U. Washington, Princeton, and CPC) Follows in footsteps of existing websites (U. Washington, Princeton, and CPC) - http://www.hydro.washington.edu/forecast/monitor/index.shtml http://www.hydro.washington.edu/forecast/monitor/index.shtml - http://hydrology.princeton.edu/forecast/ http://hydrology.princeton.edu/forecast/ - http://www.cpc.ncep.noaa.gov/soilmst/ http://www.cpc.ncep.noaa.gov/soilmst/

10. Impact of Climatology Length 1980 1997 2007 150 250 350 450 550 650 January 1st December 31st Noah LSM Total Column Moisture Climatology (mm), Southern Indiana Noah LSM Total Column Soil Moisture (mm), Southern Indiana, 1980-2007 Average Climatology Based on 10 Year Simulation Average Climatology Based on 28 Year Simulation 750 150 250 350 450 550 650 750

11. D4D3D2D1D0 D4D3D2D1D0D4D3D2D1D0D4D3D2D1D0 Impact of Climatology Length Use of longer climatology acts to decrease severity of current events by putting them into better historical context Noah LSM Total Column Soil Moisture Percentile July 1 st, 2007, Based on 10 Year Climatology Noah LSM Total Column Soil Moisture Percentile July 1 st, 2007, Based on 10 Year Climatology Noah LSM Total Column Soil Moisture Percentile July 1 st, 2007, Based on 28 Year Climatology Noah LSM Total Column Soil Moisture Percentile July 1 st, 2007, Based on 28 Year Climatology

12. Impact of Meteorological Forcing Data January 1st December 31st Noah LSM Total Column Moisture Climatology (mm), Northwest Utah Noah LSM Total Column Soil Moisture (mm), Northwest Utah, 1997-2007 1997 1999 2001 2003 2005 2007 Total Column Soil Moisture (Old Forcing) Total Column Soil Moisture (New Forcing) Average Soil Moisture Climatology (Old Forcing) Average Soil Moisture Climatology (New Forcing) 150 250 350 450 550 650 750 150 250 350 450 550 650 750

13. D4D3D2D1D0D4D3D2D1D0 D4D3D2D1D0 D4D3D2D1D0 Impact of Meteorological Forcing Data Use of new forcing data set over same 10 year time period leads to large changes in drought depiction Noah LSM Total Column Soil Moisture Percentile July 1 st, 2007, Using New Forcing Data Noah LSM Total Column Soil Moisture Percentile July 1 st, 2007, Using New Forcing Data Noah LSM Total Column Soil Moisture Percentile July 1 st, 2007, Using Old Forcing Data Noah LSM Total Column Soil Moisture Percentile July 1 st, 2007, Using Old Forcing Data

14. Impact of Model Choice January 1st December 31st LSM Total Column Moisture Climatology (mm), Northern New York Average Soil Moisture Climatology (Mosaic LSM) Average Soil Moisture Climatology (Noah LSM) 1997 1999 2001 2003 2005 2007 LSM Total Column Soil Moisture (mm), Northern New York, 1997-2007 Total Column Soil Moisture (Mosaic LSM) Total Column Soil Moisture (Noah LSM) 150 250 350 450 550 650 750 150 250 350 450 550 650 750

15. D4D3D2D1D0D4D3D2D1D0 D4D3 D2 D1D0 D4D3D2D1D0 Noah LSM Total Column Soil Moisture Percentile July 1 st, 2007, Based on 28 Year Climatology Mosaic LSM Total Column Soil Moisture Percentile July 1 st, 2007, Based on 28 Year Climatology Impact of Model Choice Choice of land surface model can greatly influence depiction of drought severity due to differences in model physics and parameterizations Noah LSM Total Column Soil Moisture Percentile July 1 st, 2007, Based on 28 Year Climatology Mosaic LSM Total Column Soil Moisture Percentile July 1 st, 2007, Based on 28 Year Climatology

16. D4D3D2D1D0D4D3D2D1D0D4D3D2D1D0 Drought Monitor Comparison Soil moisture percentiles from each LSM combined to form ensemble mean percentile map Soil moisture percentiles from each LSM combined to form ensemble mean percentile map Project will eventually use Mosaic, Noah, VIC, Sacramento, CLM3, HySSiB, and Catchment models with a variety of lineages (climate modeling, weather forecasting, hydrological) Project will eventually use Mosaic, Noah, VIC, Sacramento, CLM3, HySSiB, and Catchment models with a variety of lineages (climate modeling, weather forecasting, hydrological) Ensembles often offer more accurate depictions of drought Ensembles often offer more accurate depictions of drought Even poor depictions are informative--Large model spread indicates lack of confidence Even poor depictions are informative--Large model spread indicates lack of confidence

17. Conclusions New NLDAS LSM-based drought project underway at NASA GSFC in collaboration with NOAA/NCEP/EMC, NOAA/NCEP/CPC and Princeton University New NLDAS LSM-based drought project underway at NASA GSFC in collaboration with NOAA/NCEP/EMC, NOAA/NCEP/CPC and Princeton University Project seeks to leverage ensemble, high quality, multi- layer, spatially continuous soil moisture simulations in NLDAS framework to form a robust real-time drought monitor Project seeks to leverage ensemble, high quality, multi- layer, spatially continuous soil moisture simulations in NLDAS framework to form a robust real-time drought monitor Goals are to investigate climatology, forcing data, model, and ensemble-related issues as well as offer an effective suite of objective drought indices to drought assessment organizations such as NIDIS and the U.S. Drought Monitor Goals are to investigate climatology, forcing data, model, and ensemble-related issues as well as offer an effective suite of objective drought indices to drought assessment organizations such as NIDIS and the U.S. Drought Monitor End user input will be key to the success of this project, and all input is welcome End user input will be key to the success of this project, and all input is welcome