Developed by Thomas McKee at Colorado State University
Embraced by National Drought Mitigation Center and Western Regional Climate Center
Adapted for use in Hawaii by NWS Honolulu Forecast Office in 2000
What is the SPI?
A statistical method for assessing rainfall
More representative than using the average rainfall as a representation of what is "normal".
What's wrong with the average?
Rainfall distributions do not typically follow a Gaussian ("bellcurve") distribution.
Rainfall distributions are skewed with lower amounts being more frequent.
Result is that the average is biased toward higher values
Rainfall data are better represented by a Gamma distribution
In calculating the SPI, the observed rainfall values are first fitted to a Gamma distribution
The Gamma distribution is then transformed to a Gaussian distribution (a "bell curve").
The standardized anomaly is then computed with results having an average of zero and a standard deviation of one.
BOTTOM LINE: All of the above steps makes the SPI independent of both the location and the range of values so that different seasons and climate areas are represented on an equal basis.
Why SPI?
Palmer Drought Severity Index (PDSI) in widespread use in many parts of the U.S. BUT
PDSI not suited for high spatial variability of HI climate
PDSI depends on soil moisture and temperature data that are not routinely available in Hawaii
SPI is dependent only on rainfall data
Regularly reporting network of gages available
Results normalized - different areas can be assessed equally
Can be used to monitor wet OR dry conditions
SPI is flexible - can be applied to other types of data (e.g. streamflow, reservoir levels, etc.)
A Note on Data Sources
58 rain gages (out of about 300 sites) from 3 NWS networks
17 gages report in real-time or near real-time
41 gages manually read and report on a monthly basis
Selection based on:
Location: Tried to have several sites per climate zone
Reporting record: should be long and reliable
Real-time sites allow for SPI calculations right after the end of the month for 17 sites ("quick" report).
41 manual sites have 20 to 30 day delay in data availability. ("full" report).
SPI Product Development Process
SPI values calculated for each site from raw data
Climate zone averages determined
Text bulletin created listing SPI values for each site and for each time period (1, 2, 3, 6, 12, 18, and 24 month periods)
SPI maps created with color-coding for each climate zone.
Purple: Extremely wet (SPI at or above +2.00)
Blue: Very wet (SPI = 1.50 to 1.99)
Green: Moderately wet (SPI = 1.00 to 1.49)
Pale green: Near normal (SPI = -0.99 to 0.99)
Yellow: Moderately dry (SPI = -1.49 to -1.00)
Orange: Very dry (SPI = -1.99 to -1.50)
Red: Extremely dry (SPI at or below -2.00)
Text bulletin and SPI maps posted to NWS Honolulu website.
Short Term SPI (1-2 months)
Useful for those interested in short-term variations in rainfall
Sensitive water systems (catchment, surface water, etc.) and certain types of crops, etc.
Seasonal SPI (3 and 6 months)
Useful for certain crops and wildfire threat areas, etc.
Long-term SPI (1-year or more)
Useful for livestock, certain crops and wildfire threat areas, water supply, etc.
Sample SPI maps
Big Island
5 zones based roughly on climatology, 1 zone with no data
Maui
3 zones based roughly on climatology, 1 zone with no data