EDDI
A Powerful Tool for Early Drought Warning
Released December 2015
What is EDDI?
EDDI, which stands for Evaporative Demand Drought Index, is
a drought index that can serve as an indicator of both rapidly
evolving “flash” droughts (developing over a few weeks) and
sustained droughts (developing over months but lasting up
to years).
Why use EDDI?
EDDI has been shown to offer early warning of drought stress
relative to current operational drought indicators, such as
the US Drought Monitor (USDM) (see Figure 1). A particular
strength of EDDI is in capturing the precursor signals of water
stress at weekly to monthly timescales, which makes EDDI a
potent tool for drought preparedness at those timescales.
EDDI also uses the same classification scheme as the USDM
to define drought conditions, so it is easy to read EDDI maps.
Does EDDI work in real time?
Yes. At present, EDDI is generated every week by analyzing
a real-time atmospheric dataset. There is also an ongoing
effort to forecast EDDI based on seasonal climate-forecast
information.
What is the physical basis for EDDI?
EDDI exploits the strong physical relationship between
evaporative demand (
E
0
) and actual loss of water from the
land surface through evapotranspiration.
E
0
is the “thirst of the
atmosphere,” estimated by the amount of water that would
evaporate from the soil and be transpired by plants if the
soil were well watered. EDDI measures the signal of drought
using information on the rapidly evolving (daily) conditions
of the atmosphere to estimate their impact on land-surface
moisture, and vice versa. EDDI’s effectiveness in reflecting the
moisture conditions on the land surface is based on feedbacks
between the atmosphere and land that are particularly strong
during the warm season, when drought is of greatest concern.
EDDI is sensitive to two distinct land-surface atmosphere
interactions: (i) increased
E
0
drives increased evapo-
Figure 1
Development of a flash drought in the Midwest in 2012. The
2-week EDDI (right) is compared at 5-week intervals to the
US Drought Monitor (USDM) (left). EDDI captures the severe
drought condition two months ahead of the USDM. Image:
Mike Hobbins.
transpiration until the available soil moisture becomes
limiting, potentially leading to flash droughts; and (ii) as
surface water becomes increasingly scarce in sustained
droughts, evapotranspiration declines, which leads to higher
air temperature and lower humidity, and thereby increases
E
0
.
Green River, Wyoming. Photo: K. Miller, USGS.
EDDI
A Powerful Tool for Early Drought Warning
Is the calculation of EDDI sensitive
to land-surface type?
No, EDDI is based on relative changes in
E
0
, so it
is not sensitive to land-surface type and is a valid
drought indicator for all regions.
How is EDDI calculated?
EDDI is a measure of the departure of
E
0
aggregated across a time-window of interest
relative to historical conditions (from a 30-
year climatology).
E
0
is calculated as reference
evapotranspiration based on the FAO-56
Penman-Monteith formulation. At each point in
space, the rank of aggregated
E
0
relative to its
climatology is converted to a percentile, which
is then assigned to different drought categories
– e.g., ED0, ED1, etc. – that are equivalent to the
categories used by the USDM. The time-window
of interest could be specific weeks or months in
any given year, up to the present. For example, a
2-week EDDI will aggregate two weeks of
E
0
and
estimate its departure relative to the historical
aggregation of
E
0
for those two weeks. Figure 2
shows an example of 2-week EDDI for the Wind
River Indian Reservation in central Wyoming
between May 26, 2015, and September 29, 2015.
What time-window information on
EDDI is most appropriate?
The optimal time-window for EDDI will be user-
and sector-dependent. For example, an irrigator
may be interested in short-term EDDI – say
across a 2-week window – to track and respond
to weather-scale changes, whereas a reservoir
operator more interested in interseasonal
variations in snowpack may find more utility
in a 6-month EDDI that examines the behavior
across the snow accumulation and snowmelt
periods. Given the natural and physical linkages
between
E
0
and wildfire risk, work is ongoing to
establish the optimal timescales for fire-weather
prediction. Indeed, the US Forest Service Rocky
Mountain Research Station uses 1-month EDDI
in their seasonal forecasting of the numbers
of large fires and of fire-suppression costs. For
other cases, consideration of EDDI information
at multiple timescales would be useful.
Where can I get EDDI data and maps?
EDDI is an experimental product. To receive
EDDI data and maps, please contact Mike
Hobbins (
mike.hobbins@noaa.gov
). Currently,
EDDI maps for the Rocky Mountain region are
being presented in the weekly NIDIS Upper Colorado River Basin Drought
and Water Assessment (
http://climate.colostate.edu/~drought
). In the near
future, we hope to make EDDI data available through the WWA Climate
Dashboards and, looking further out, to have EDDI data distributed
nationally by the National Weather Service.
Figure 2
Drought development in the Wind River Indian Reservation across the
irrigation season, as observed by a 2-week EDDI at 2-week intervals. Drying
did not appear in the US Drought Monitor until September 29 (blue-outlined
map at bottom right). Image: Mike Hobbins.
References
Hobbins MT, Wood A, McEvoy D, Huntington J, Morton C, Anderson M, Hain C, and
Verdin, The Evaporative Demand Drought Index: Part I - Linking drought evolution
to variations in evaporative demand. Journal of Hydrometeorology (In revision).
McEvoy D, Huntington J, Hobbins MT, Wood A, Morton C, Anderson M, Hain C, and
Verdin J, The Evaporative Demand Drought Index: Part II - CONUS-wide assessment
against common drought indicators. Journal of Hydrometeorology (In revision).
Hobbins, MT (2014), Measuring the Atmosphere’s Thirst. Dry Times: National
Integrated Drought Information System Newsletter 4 (Apr. 2014): 14-15.
http://www.drought.gov/media/pgfiles/NIDIS-Newsletter-April-2014.pdf
Authors
Imtiaz Rangwala
1,2,3,4
(
Imtiaz.Rangwala@noaa.gov
),
Mike Hobbins
1,2,5
,
Joe Barsugli
1,2,3,4
and Candida Dewes
1,2,3,4
1 Cooperative Institute for Research in Environmental Sciences (CIRES), 2 NOAA
ESRL Physical Sciences Division (PSD), 3 North Central Climate Science Center
(NCCSC), 4 Western Water Assessment (WWA), 5 National Integrated Drought
Information System (NIDIS)
Acknowledgements
Ami Nacu-Schmidt (CIRES/WWA) for design and layout of this document.