North American Drought in the 21st Century Project OverviewDennis P. Lettenmaier

University of WashingtonEric F. Wood

Princeton UniversityGordon Bonan

National Center for Atmospheric Research

Climate Change Prediction Program PI Meeting

Seattle

October 18, 2004

Central science question:

What is the susceptibility of the continental U.S. to drought over the next century, and what role is anthropogenic warming likely to play in U.S. drought susceptibility?

Drought: Most Costly Natural Disaster

• Average damages in U.S.: $6-8 billion/year

• Major impacts:agriculturalmunicipalindustrial environmental

“We propose to use the Community Climate System Model (CCSM), in conjunction with a modified version of the Community Land Model (CLM) that will incorporate a more realistic representation of land surface hydrology, to evaluate the susceptibility of the U.S. to drought over the next century.”

Subsidiary questions:

a) What have been the space-time signatures of 20th century drought on precipitation, soil moisture, and streamflow, and how might those change in the 21st century?

b) What is the role of climate-vegetation feedbacks in exacerbating or ameliorating North American drought severity and intensity?

c) Can useful estimates of drought recovery probabilities be developed based on off-line and or coupled ensemble climate prediction methods?

Tasks:Task 1: Improve land surface hydrology representation in CLM. (By

“transplanting” aspects of the runoff generation, soil column representation, and possibly the snow model from the Variable Infiltration Capacity (VIC) model into CLM)

Task 2: Perform offline tests of upgraded version of CLM. (Specifically using updated PILPS-2c data for the Arkansas-Red River basin, and other PILPS data)

Task 3: Evaluate performance of upgraded model in climate simulations. (Focus on the period 1950-present, for which CCSM runs with observed SSTs can be compared with off-line land surface simulations)

Task 4: Simulate and evaluate projected drought characteristics. (To be based on transient climate simulations for the period 1870-2100)

Task 5: Drought recovery evaluation. (Explore the possibility of using coupled and/or off-line simulations to evaluate alternate methods of estimating drought recovery.)

Task 6: High resolution simulations for the U.S. (Develop a prototype high-resolution version of the CLM that runs on a finer spatial grid than the current CLM – perhaps 1/8 degree LDAS grid for the U.S.)

Hydrologic Model

Variable Infiltration Capacity (VIC) Model Features:

•Developed over 10+ years•Energy and water budget closure at each time step

•Multiple vegetation classes in each cell

•Sub-grid elevation band definition (for snow)

•Subgrid infiltration/runoff variability

•History of application to many large (continental) river basins

Macroscale modeling approach (“top down”)

1 Northwest 5 Rio Grande 10 Upper Mississippi2 California 6 Missouri 11 Lower Mississippi3 Great Basin 7 Arkansas-Red 12 Ohio4 Colorado 8 Gulf 13 East Coast

9 Great Lakes

Comparison with Illinois Soil Moisture

19 observing stations are compared to the 17 1/8º modeled grid cells that contain the observation points.

Persistence

Moisture Level

Moisture Flux

Variability

Obs.Model

Maurer et al., 2002

SWE, 1981

Drought Definitions

• Meteorological Drought– Precipitation and

Temperature• Agricultural Drought

– Soil Moisture• Hydrologic Drought

– Streamflow/Runoff

http://nm.water.usgs.gov/drought/photos.htm

How can we use information from long term hydrologic model

simulations to synthesize the following drought characteristics:

severity, intensity, extent, and duration?

Severity-Area-Duration Analysis

• Based on the Depth-Area-Duration technique from probable maximum precipitation analysis

• Replace depth with measure of drought severity

• S=(1-ΣP/t) – S=severity, ΣP = cumulative percentile (soil

moisture or runoff), t = duration

How do we define drought extent?

SAD Construction

1. Rank cells by severity & identify potential drought centers

2. Search 3x3 neighborhood of drought center

3. Average severities & add areas4. Output severity and area at

specified area intervals5. Compare the severity at ~25,000

km2 for each potential drought center and select center with maximum severity

Modified from WMO (1960) computational method of DAD analysis

1)

2)

3)

MethodologyVIC model outputTotal column soil moistureRunoff

Weibull percentiles

Threshold20th percentile and lower soil moisture30th percentile and lower runoff

Spatial contiguity

Initial drought classification

Temporal contiguity

Final drought and subdrought classification

Severity-Area-Duration

SAD curves for each event

Highest severities

Envelope curve for each duration

Soil Moisture

3-month

6-month

1 year

2 years4 years

8 years

Dec 1976 to Feb 1977

Feb 2002 to Feb 2003

Feb 1930 to Feb 1937

Oct 1951 to Oct 1956

Runoff

3-month

6-month1 year2 years

4 years8 years

Feb 2001 to Feb 2002

Sep 1976 to Nov 1976

Jan 1934 to Mar 1934

Sep 1953 to Sep 1956

Runoff vs. Soil moisture

• Lower severities at higher areas and longer durations for runoff

• Current drought more dominant in runoff due to sensitivity to precipitation deficit?

Summary• Proposal is to merge best aspects of CLM and

VIC, with emphasis on producing a land surface scheme that is capable of reproducing observed land surface hydrologic dynamics of major river basins

• Most likely aspects of VIC to be integrated into CLM are runoff production parameterizations (variable infiltration capacity curve; baseflow parameterization) and perhaps snow model; energetics and vegetation parameterizations will likely be retained

• Drought severity-area-duration analysis (SAD) will be the mechanism for investigation of implications of future climate for drought (through both off-line and coupled applications)