Description of Standardized Precipitation Index (SPI) and Canadian Evaluation of SPI in Diverse Climates

1
Description of Standardized Precipitation Index
(SPI) and Canadian Evaluation of SPI in Diverse
Climates
Richard R. Heim Jr. NOAA/NESDIS/National
Climatic Data Center Asheville, North Carolina,
USA E.G. (Ted) OBrien Environment
Canada Regina, Saskatchewan, Canada
2
Description of the Standardized Precipitation
Index (SPI)
Richard R. Heim Jr. NOAA/NESDIS/National
Climatic Data Center Asheville, North Carolina,
USA material from Mark Svoboda, National
Drought Mitigation Center Global Drought
Assessment Workshop Asheville, NC, USA 21 April
2010
3
Applying the Standardized Precipitation Index as
a Drought Indicator
Mark Svoboda, Climatologist Monitoring Program
Area Leader National Drought Mitigation
Center University of Nebraska-Lincoln
Inter-Regional Workshop on Indices and Early
Warning Systems for Drought Lincoln, NE December
8-11, 2009
4
Characteristics of the Standardized Precipitation
Index (SPI)

• Developed by McKee et al. in 1993
• Simple index--precipitation is the only parameter
  (probability of observed precipitation
  transformed into an index)
• Being used in research or operational mode in
  over 60 countries
• Multiple time scales allow for temporal
  flexibility in evaluation of precipitation
  conditions and water supply

5
How it Works

• It is NOT simply the difference of precipitation
  from the mean divided by the standard deviation
• Precipitation is normalized using a probability
  distribution function so that values of SPI are
  actually seen as standard deviations from the
  median
• A normalized distribution allows for estimating
  both dry and wet periods
• Accumulated values can be used to analyze drought
  severity (magnitude)

6
How it Works

• Need 30 years of continuous monthly precipitation
  data (the longer the better)
• SPI time scale intervals shorter than 1 month and
  longer than 24 months may be unreliable
• Is spatially invariant in its interpretation
• Probability based (probability of observed
  precipitation transformed into an index) nature
  is well suited to risk management

7
SPI Methodology

• The SPI calculation for any location is based on
  the long-term precipitation record for a desired
  period. This long-term record is fitted to a
  probability distribution, which is then
  transformed into a normal distribution so that
  the mean SPI for the location and desired period
  is zero (Edwards and McKee, 1997)
• Positive SPI values indicate greater than median
  precipitation, and negative values indicate less
  than median precipitation
• Because the SPI is normalized, wetter and drier
  climates can be represented in the same way, and
  wet periods can also be monitored using the SPI.

8
SPI Methodology

• Overview The SPI is an index based on the
  probability of precipitation for any time scale.
• Who uses it Many drought planners appreciate the
  SPIs versatility. U.S. North American Drought
  Monitors, U.S. State Drought Plans. Over 60
  countries.
• Pros The SPI can be computed for different time
  scales
• can provide early warning of drought and help
  assess drought severity
• less complex than the Palmer.
• One number/has historical context
• Cons Based on Precipitation only
• no Temp, no ET.
• Values based on preliminary data may change.
• Not as applicable to CC analysis

9
SPI Methodology

• The SPI was designed to quantify the
  precipitation deficit for multiple time scales
• These time scales reflect the impact of drought
  on the availability of the different water
  resources
• Soil moisture conditions respond to
  precipitation anomalies on a relatively short
  scale. Groundwater, streamflow, and reservoir
  storage reflect the longer-term precipitation
  anomalies
• For these reasons, McKee et al. (1993) originally
  calculated the SPI for 3, 6,12, 24, and
  48month time scales.

10
SPI data used in the U.S. Drought Monitor

• D0 Abnormally Dry SPI value of -0.5 to -0.7
• D1 Moderate Drought -0.8 to -1.2
• D2 Severe Drought -1.3 to
  -1.5
• D3 Extreme Drought -1.6 to -1.9
• D4 Exceptional Drought -2.0 or less
• NDMC Daily Gridded SPI Product

11
Probability of Recurrence
SPI Category of times in 100 yrs. Severity of event
0 to -0.99 Mild dryness 33 1 in 3 yrs.
-1.00 to -1.49 Moderate dryness 10 1 in 10 yrs.
-1.5 to -1.99 Severe dryness 5 1 in 20 yrs.
lt -2.0 Extreme dryness 2.5 1 in 50 yrs.
12
NDMC Distribution of SPI
Provided to over 60 countries 150
scientists Over 50 visiting scientists
13
Considerations

• Different probability functions used to fit the
  SPI will result in different values
• Different periods of record used to standardize
  the SPI will result in different values
• The same probability distribution functions and
  period of record need to be used for all stations
  to ensure spatial comparability between stations
  between countries
• The various time scales for the SPI reflect
  different forcings and types of drought
  (meteorological vs. hydrological) this may vary
  with location and season

14
Thank you! Questions? For SPI monthly code (PC
or UNIX based), E-mail Mark Svoboda at
(msvoboda2_at_unl.edu) Incomplete Gamma pdf (SPI
code also available at NCDC Richard.Heim_at_noaa.gov
) Pearson III pdf
15
Canadian Preliminary Evaluation of SPI in Diverse
Climates
E. G. (Ted) OBrien and Jennifer Stroich, 2004

• Presented by E. G. (Ted) OBrien, Environment
  Canada, Meteorological Services of Canada

16
Canadian Test Sites
Swift Current, Saskatchewan
Guelph, Ontario
Kentville, Nova Scotia (Annapolis Valley)
17
Swift Current Streamflow Findings (1960 2001)

• SPIs for Swift Current compared to monthly
  streamflow - Swift Current Creek near Leinan
  (regulated) Gross Drainage Area 3730 km2
• Tested against 1-,2-,3-,6-,12-,24- month SPIs -
  ECORCs Swift Current CDA
• SPIs under 12 months correlate best with May to
  August streamflow
• 2- to 3-month SPIs strong correlations during
  summer months
• 6- month SPI strongest correlations during summer
  months
• Beyond 6-month SPI significant correlations
  decline

18
(No Transcript)
19
Guelph Streamflow Findings (1970 2001)

• SPIs for Guelph compared to monthly streamflow
  Grand River at Galt (regulated) - Gross Drainage
  Area 3520 km2
• Tested against 1-,2-,3-,6-,12-,24- month SPIs
  from ECORCs Waterloo-Wellington A and Fergus
  Shand Dam
• 1- to 2-month SPI - correlations sporadic
• 3- month SPI correlations - significant all times
  of year, exception April
• Beyond 3- month SPI the number of significant
  correlations declines

20
Kentville Streamflow Findings (1960 1995)

• SPIs for Kentville compared to monthly streamflow
  at the Annapolis River at Wilmot (regulated)
  Gross Drainage Area 546 km2
• Tested against 1-,2-,3-,6-, 12-, 24- month SPIs
  from ECORCs Kentville CDA
• 2- and 3- month SPI correlations strongest
• Beyond 6- month SPI number of significant
  correlations decline

21
Swift Current Yield Findings (1960 2001)

• SPI Swift Current CDA and AAFC spring wheat
  research trial at Swift Current CDA (fallow-wheat
  rotation)
• Tested against 1-,2-,3-,6-,12-,24- month SPIs
  from ECORCs Swift Current CDA
• Spring wheat yields correlated best with July
  SPI.
• July 2-year SPI (r 0.750)
• July 3-month SPI (r 0.712)

22
(No Transcript)
23
Guelph Yield Findings (1970 2001)

• SPIs compared to AGRICORP Wellington county crop
  yield data
• Tested 1-, 2-, 3-, 6-, 12-mth, 2-, 3-, 4-, 5-yr
  SPI calculated from ECORCs Waterloo-Wellington A
  and Fergus Shand Dam
• 2- month SPI strongest correlations
• Correlations are negative or positive
• Investigated years of above average, average and
  below average precipitation
• SPI or percent of average precipitation do not
  appear to apply to yields
• 1989 dry July September resulted in yields
  below the 10th percentile for canola and spring
  grain
• Temperature strong positive correlations exist

24
Kentville Yield Findings (1960 1995)

• SPIs for Kentville compared to crop yields.
• Tested against 1-,2-,3-,6-, 12-, 24- month SPIs
  from ECORCs Kentville CDA
• No correlations exist with apple yield data
• Information obtained for Kings County indicated
  that most of the potatoes, beans and peas and 90
  of wheat grown in the province is grown here
• Wheat yields obtained at the provincial level -
  results indicate statistically significant
  correlations - precipitation accounts for 20
  of the variability in yield

25
Conflicts among indices

• Moderate drought (D1) using the percent of normal
  precipitation while the SPI values classified an
  extreme drought (D3).
• Example 1989 Guelph, ON
• D1 (moderate drought) using percent of normal
  precipitation from July, August and September
• D4 (exceptional drought) using the September
  3-month SPI
• For this reason we recommend using precipitation
  percentiles

26
Findings continued

• Dry spells of 4 8 weeks duration and
  temperature extremes
• Hydrologic drought (at a broad scale) use of
  monthly or seasonal streamflow volume percentiles
• Statistical correlations with precipitation
  explain very little (less than 25) of observed
  yield variations in more humid regions