Monitoring Drought in the

1
Monitoring Drought in the 21st Century A
Resolution Revolution
Mark Svoboda National Drought Mitigation
Center International Drought Information
Center University of Nebraska-Lincoln
2
Why Monitor Drought?

• Drought is a Normal Part of the Climatic Cycle
• Drought Impacts are Significant Widespread
• Many Economic Sectors Affected
• Drought is Expensive
• Since 1980, major droughts and heat waves within
  the U.S. alone have resulted in costs exceeding
  100 billion dollars (NOAA/NCDC)
• Average annual losses 6-8 billion (FEMA 95)

3
An integrated climate monitoring system needs to

• be comprehensive in scope (coupling climate, soil
  and water data)
• incorporate local and regional scale data
• use the best available (multiple) indices and
  triggering tools
• link index values or thresholds to impacts
• be flexible, incorporating the needs of users

4
The Importance of a Drought EWS

• allows for early drought detection
• allows for proactive (mitigation) and reactive
  (emergency) responses
• triggers actions within a drought plan
• Bottom line?provides information for decision
  support

5
Components of a Drought EWS

• timely data and timely acquisition
• synthesis/analysis of data used to trigger set
  actions within a plan
• efficient dissemination or delivery system (WWW,
  media, extension)

6
Importance of Drought Indices

• Simplify complex relationships and provide a good
  communication tool for diverse audiences
• Quantitative assessment of anomalous climatic
  conditions
• Intensity
• Duration
• Spatial extent
• Historical reference (probability of recurrence)
• Planning and design applications

7
Triggers thresholds determining specific, timely
actions by decision makers. Link impacts to index
or indicator values.
Triggers need to be

• appropriate
• consistent with impacts
• adaptable

8
Considerations for Selecting a Specific Trigger
or Index

• Is the information readily available?
• Is the information likely to remain available
  over time?
• Can an index/trigger be calculated in a timely
  manner?
• Can the index/trigger be meaningfully correlated
  to actual conditions (impacts)?

9
Approaches to Drought Monitoring (assessment)

• Single index or parameter
• Multiple indices or parameters
• Composite index

10
The U.S. Drought Monitor
Since 1999, NOAA (CPC and NCDC), USDA, and the
NDMC have produced a composite drought map -- the
U.S. Drought Monitor -- each week with input from
numerous federal and non-federal agencies
11
U.S. Drought Monitor Map
Drought Intensity Categories
D0 Abnormally Dry
D1 Drought Moderate
D2 Drought Severe
D3 Drought Extreme
D4 Drought Exceptional
12
U.S. Drought Monitor

• Several key and ancillary indicators
• Attempts to capture conditions across
  widespectrum of drought conditions
• Must Address
• No single definitionof drought
• Integrates manyindicators
• Now creating in
• ARC GIS

13
http//drought.unl.edu/dm
14
(No Transcript)
15
Continental-Scale Indicators

• Standardizing Period 1951-2001 (for PCTPCP,
  SPI, PDI)
• Guidance for drought category boundaries across
  international borders

16
Is There a Need for a Water Resources Monitor?

• Mark Svoboda
• National Drought Mitigation Center

With Contributions From Harry Lins, USGS Phil
Pasteris, USDA/NRCS Frank Richards, NOAA
17
Water Monitor Concept

• A proposed partnership between the USGS, NOAA,
  USDA and the NDMC
• Idea hatched in spring 2003 as a potential answer
  to the debate of 1 vs. 2 DM maps (short- and
  long-term drought) and as a better tool for the
  DM authors
• A consolidation of indices and hydrological
  indicators by basin or hydrological unit
• The Water Resources Monitor portal (maintained
  by USGS) would provide for a general hydro
  assessment of water resources in the U.S. to
  compliment the Drought Monitor

18
Water Resources Monitor
19
Water Resources Monitor
20
BASIN WATER INDEX (BWI)

• Components precipitation (antecedent precip
  index (API) used), snowpack, streamflow,
  reservoir storage, well observations
• Based on hydrological units
• National application potential using a modified
  SWSI approach
• Flexible to include any combination of collected
  variables
• Overall index values can aid DM authors or other
  users in monitoring drought
• Non-exceedance values can be computed for the
  individual components or for the aggregate
  variable

21
Basin Water Index Provides for a
nationwide (potentially) water availability index
that minimizes the documented limitations of the
SWSI (Still under development)
Water and Climate Center
22
National Agricultural Decision Support System

• Allows for weekly table/map output of the
• SPI, PDSI and Newhall values
• 1948-present (base 61-90 for distribution)
• HPRCC/ACIS SHEF formatted real-time data from
  COOP and first order sites

23
http//nadss.unl.edu
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
Standardized Precipitation Index (SPI) by
Climate Division (above), and by 0.25º grid
(below)
U.S. Department of the Interior U.S. Geological
Survey EROS Data Center
29
Assessing and Monitoring Drought with Climate,
Satellite, and Other Geophysical Data Sets Using
Data Mining Techniques

• T. Tadesse, J. Brown, and M. Hayes
• National Drought Mitigation Center, University
  of Nebraska-Lincoln
• USGS/EROS Data Center, Sioux Falls, SD

30
Integrating Satellite and Climate Data

• Percent of Average Seasonal Greenness highlights
  areas with lower than average vegetation
  condition
• Can be caused by drought, flooding, late
  greennup, land cover conversion, etc.
• Climate-based drought indicators enable
  interpretation of the satellite data anomalies

31
Methodology
Model Input
VegDRI (Vegetation Drought Response Index
Avg Seasonal Greenness
32
Data Mining

• Data mining is a method that brings techniques
  from machine learning, pattern recognition,
  statistics, databases, and visualization together
  to address the issue of information extraction
  from large databases (Cabena et al., 1998)
• Data mining tools can answer questions that
  traditionally were too time-consuming to resolve
• high computing, fast data access, large storage
  of data
• Search databases for hidden patterns and find
  predictive information that experts may miss
  because it lies outside their expectations
  (Thearling, 2001)

33
Prototype Veg-DRI May 30, 2002
34
Prototype Veg-DRI July 25, 2002
35
Prototype Veg-DRI September 5, 2002
36
July 11, 2002
Drought Monitor July 9, 2002
37
A County Story
Hand County, South Dakota
38
Hand County, SD

• Total area 3,705 km2 (9 million acres)
• Top Five Commodities (from 1997 Ag census)
• Cattle and calves
• Wheat
• Corn (for grain)
• All other grains
• Hogs and pigs

39
Land Cover (from National Land Cover
Database) Pink Row Crops Burgundy Small
Grains Tan Grassland Yellow Pasture/Hay
40
Hand County, SD
May 30, 2002
July 25, 2002
September 5, 2002
47 of county is normal 53 of county is in
drought
41
Hand County, SD
July 25, 2002
Land Cover
168,000 acres of grassland impacted by
drought 213,000 acres of row crops impacted by
drought
42
(No Transcript)
43
Conclusion

• Better resolution (1 km2 pixel) drought response
  maps were produced as opposed to drought maps
  that were traditionally made at the county and
  climate division level
• Better information on vegetation stress
  particularly for pasture and rangelands
• Integrates both satellite and climate information
  at the same time
• Rule-based models can be used in predicting the
  agricultural condition of drought up to six weeks
  in advance with relatively higher accuracy
• Gives prospects of timely information (near-real
  time web-delivery) for decision makers
• Caveats should be used as complementary
  information until it is fully operational

44
Future Work

• Progress
• Web-based delivery of near real-time data and
  maps (http//gisdata.usgs.net/website/Drought_Moni
  toring/viewer.asp)
• Challenges
• Extending from Central Plains to conterminous
  U.S.
• Transitioning from research and development stage
  to operational system

45
General Views on Drought Forecasting

• Easiest during strong climate fluctuations (La
  Nina, El Nino, etc.)
• A combination of art and science, needing to
  skillfully combine short-term and long-term
  signals
• Forecasts are slowly improving as they gain
  experience in using both dynamic and statistical
  forecasts as well as climatology
• Much more research needed to determine how the
  atmosphere, oceans, and continents talk to each
  other

46
CPC Was Among the First to Produce Seasonal
Drought Outlooks
(from Climate Prediction Center)
47
For the CPC Seasonal Drought Outlooks.

• CPC uses historical information (ENSO composites,
  natural and constructed analogues, drought
  probabilities) to assess the odds of drought
  continuing or dissipating
• Also incorporate forecast information for all
  appropriate time periods (short, medium, and
  long-range) and consider the evolution of
  hemispheric circulation patterns

48
Climate Indicators that Correlate with U.S.
Drought

• El Niño/La Niña
• PDO
• Solar Cycle
• Arctic Oscillation
• Pacific Warm Pool SSTs
• NH and Global Temperatures

49
Simplified Forecast Procedures

• Start out with Drought Monitor D1 areas
• Forecast persist or intensify where seasonal
  outlook shows below-normal precipitation
• Forecast improvement where seasonal outlook shows
  above-normal precipitation
• Use Palmer Drought projections, analogues,
  composites, and other tools where seasonal
  outlook shows equal chances

50
ENSO Composites
Constructed Analogue Soil Model
Palmer Drought Projections
Medium-Range Forecast
51
Future Drought Monitoring Challenges

• Maintain the momentum of the current Resolution
  Revolution
• ACIS (Mesonets)
• NOAA/USGS gridded mapping/COOP Modernization
• Satellite/Radar
• NADSS
• USGS
• Others
• Develop a comprehensive water resources
  monitoring tool (The Water Resources Monitor??)
• Enhancing our nations observed soil moisture
  networks (SCAN or Mesonet operated)

52
Future Drought Monitoring Challenges

• Impact collection/reporting tools
• Improved drought prediction
• Robust IMS/GIS query/analysis potential for
  Drought Monitor and other products
• Research, research, research

53
Critical Observations
1) No single parameter is used solely in
determining appropriate actions 2) Instead,
different thresholds from different combinations
of inputs is the best way to approach monitoring
and triggers 3) Decision making (or triggers)
based on quantitative values are supported
favorably and are better understood
54
http//drought.unl.edu
55
Questions? Email me at msvoboda2_at_unl.edu Or
call [phone]
56
(No Transcript)
57
Potential Monitoring System Products and Reports

• Historical analysis (climatology, impacts,
  magnitude, frequency)
• Operational assessment (coop network data, SPI
  and other indices, automated networks, satellite
  and soil moisture data)
• Predictions/Projections (SPI and other indices,
  soil moisture, streamflow, seasonal forecasts,
  SSTs)

58
Why the Recent Interest in Drought in the U.S.?

• Single and multi-year severe droughts
• intensity and duration
• western and eastern U.S.
• Spatial extent40 to 50 of U.S. in 2002-03
• Complexity of impacts ? Vulnerability
• Agriculture, energy, transportation, urban water
  supply, recreation/tourism, fires, environmental,
  social
• Conflicts between water users
• Water restrictions (agricultural and urban)
• Trend toward drought mitigation planning
• Media coverage

59
Recent Drought Losses in the U.S.
1988 39.2 billion nationwide 1993 1 billion
across the Southeast 1996 10 billion across
the Southwest 1998 6-8 billion across the
South 1999 1 billion along the East
Coast 2000 1 billion each in Nebraska,
Oklahoma, Texas, and Georgia 2002
gt20 billion nationwide?? 2003 billion ????
Average annual losses 6-8 billion (FEMA)
60
Components of a Drought Plan

• monitoring, early warning, and prediction
• risk and impact assessment
• mitigation and response

61
Questions addressed by monitoring

• Analyze recent eventshow did we get here?
• Place current situation in a historical
  contexthow rare is this event?
• What is the forecast and how reliable is it?
• What would it take to end the drought event?
• How can we communicate this information to
  decision makers to encourage positive action?

62
(No Transcript)
63
Key Variables For Monitoring Drought

• climate data
• soil moisture
• stream flow
• ground water
• reservoir and lake levels
• snow pack
• short, medium, and long range forecasts
• vegetation health/stress and fire danger

64
Indicators for The West
65
Original Objectives

• Fujita-like scale
• NOT a forecast!
• Identify impacts (A, H)
• Assessment of current conditions
• Incorporate local expert input
• Be as objective as possible

66
Drought Severity Classifications
Indices used primarily during the snow season
and in the West include the River Basin Snow
Water Content, River Basin Average Precipitation
and SWSI
67
Objective Blends
Useful for showing situations and areas having
similar trends or opposite trends in moisture
conditions
68
Objective Blends

• Short-Term Blend
• 35 Palmer Z Index
• 25 3-Month Precip.
• 20 1-Month Precip.
• 13 CPC Soil Model
• 7 Palmer Drought
• Index

69
Objective Blends

• Long-Term Blend
• 25 Palmer Hydrological Index
• 20 24-Month Precip.
• 20 12-Month Precip.
• 15 6-Month Precip.
• 10 60-Month Precip.
• 10 CPC Soil Model

70
A Possible Approach

• Start w/ streamflow data
• Supplement as best as possible w/ other data
• Base analysis on HUCs
• Use GIS/IMS as key tools
• Map depicts where impact occurs?
• Value added from experts/subjective assessment
  and interpretation of the objective factors
• Maintain web site and open listserver like the DM

71
Next Steps

• Pitch importance/need for an integrated water
  resource assessment portal
• Develop prototype product/website by January 2004
• Get input from you!

72
The Importance of Local Expert Input

• The National Centers can produce a variety of
  input indicator products(e.g., CPC station dot
  map)
• These give us The Big Picture

73
The Importance of Local Expert Input

• The U.S. Drought Monitor Team Relies on Field
  Observation Feedback from the Local Experts for
  Impacts Information Ground Truth
• Listserver (140-150 Participants 2/3 Federal,
  1/3 State/Univ.)

Local NWS USDA/NRCS Offices State Climate
Offices State Drought Task Forces Regional
Climate Centers
Midwest Regional Climate Center
74
The Importance of Local Expert Input
High PlainsRegionalClimate Center
75
The Importance of Local Expert Input
Colorado Climate Center
Western Regional Climate Center
76
The Importance of Local Expert Input
Montana StateDrought Advisory Committee
Oregon State Climatologist Office
77
Questions

• Is there a need for a hydro-oriented companion to
  the Drought Monitor?
• Whos the audience?
• Whats needed to make it happen?

78
Observed real-time data are essential

• for timely drought assessments (real-time and
  historical)
• for increased spatial and temporal resolution
• as input for generating many climate
  products/forecasts
• ground truthing of soil moisture (and other)
  models
• ground truthing of radar precipitation
  estimates
• getting information to decision makers when they
  need it.i.e. yesterday!
• filling in data sparse areas

79
Mesonet Sites Approx. 1,000 in the U.S.
80
Next Steps

• Incorporate elevation factor (SNOTEL)
• Integrate TD3206 data for pre-1948 analysis
• Incorporate projected SPI maps based on
• .80/.50/.20 for decision makers
• Implement Pearson III in the code
• Extend out to 104 weeks (2 years)
• More exposure analysis research

81
Why should we integrate satellite and climate
data ?

• Percent of Average Seasonal Greenness highlights
  areas with lower than average vegetation
  condition
• Vegetation stress can be caused by drought,
  flooding, late Greenup, land cover conversion,
  etc.
• Climate-based drought indicators will enable
  interpretation of the satellite data anomalies
• Geophysical parameters such as land cover types
  help identify areas with vegetation stress due to
  drought
• Data mining techniques used to create a model
• Better drought impact map can be produced to
  monitor drought

82
Experimental Results

• The total of 224 stations in Nebraska and South
  Dakota were used
• Cubist data mining software was used to generate
  a rule-based model
• Data Period 1989 to 2002 (14 years)
• Climatic variables bi-weekly SPI and PDSI
• Satellite geophysical variables
• percent of irrigated areas
• ecological type
• land cover type
• satellite-derived start of the vegetation season
• available water capacity of the soil
• satellite-derived percent of vegetation

83
What We Need for Better Forecasts

• Better depiction of current drought areas through
  improved monitoring capabilities
• More accurate forecasts of temperature and
  precipitation, especially for seasonal periods
• Better ways to incorporate forecasts of
  temperature and rainfall into forecasts of
  drought indices (Palmer, soil moisture,
  streamflows)
• Greater understanding of surface-air feedback
  processes through statistical and dynamic
  modeling improvements
• Better ways to depict the probabilities of
  drought worsening or improving

84
(No Transcript)
85
(No Transcript)
86
(No Transcript)
87
(No Transcript)
88
Observational/Data Requirements

• Primary
• surface water
• reservoir
• groundwater
• Others
• snow
• soil moisture
• water supply forecasts
• PRISM to address elevation