Precipitation, as a variable, can be transformed into several types of indices:
1. Percent of normal can analyze a single region or a single season, yet it is easily misunderstood and gives different values depending on the location and time period. Further, mean precipitation (the average amount) usually differs from median precipitation (the amount exceeded 50% of the time) because precipitation tends to be skewed rather than normally distributed. For a positively skewed precipitation distribution, the median is less than the mean, so below-normal (below-average) precipitation is more likely than above-normal precipitation. For instance, in Melbourne, Australia, median precipitation for February is 32.4 mm, but this is only 68.6% of "normal" when compared to the mean (47.2 mm) (AU-CBM, 2003). Using percent of normal can make it difficult to link a value of a departure with a specific impact occurring as a result of the departure, and thus to design appropriate drought mitigation and response measures (Willeke et al., 1994).
2. Deciles (Gibbs and Maher, 1967) can address limitations of the percent of normal approach. The long-term precipitation record is divided into tenths of percentiles, called deciles: the lowest 20% is much below normal; next lowest 20% is below normal; middle 20% is near normal; next highest 20% is above normal; and highest 20% is much above normal. The deciles method was selected over the PDSI for the Australian Drought Watch System for simplicity, con sistency, and understandability (Smith et al., 1993). One challenge, though, is that a long climatological record with consistent observation stations is needed to calculate the deciles accurately. Also, deciles can be difficult to apply if officials and the public are not familiar with the system.
3. Standardized Precipitation Index (SPI), developed by McKee et al. (1993), quantifies precipitation deficit for multiple timescales, such as for 3-, 6-, 9-, and 12-month prior periods, relative to those same months historically. These different timescales are designed to reflect the impacts of precipitation deficits on different water resources. For instance, soil moisture conditions respond to precipitation anomalies on a relatively short scale, whereas groundwater, stream-flow, and reservoir storage reflect longer term precipitation anomalies.
The SPI relies on a long-term precipitation record, typically at least 30 years, for a desired region, such as a climate division. This record is fitted to a probability distribution, such as the gamma distribution or Pearson III, so that a percentile on the fitted distribution corresponds to the same percentile on a gaussian distribution (Panofsky and Brier, 1958). That percentile is then associated with a Z score for the standard gaussian distribution, and the Z score is the value of the SPI.
The categories of the SPI, according to McKee et al. (1993), are as follows:
SPI Values Drought Category Cumulative Frequency
0 to -0.99 Mild drought 16-50%
-2.00 or less Extreme drought <2.3%
One advantage of the SPI is that it is standardized, so its values represent the same probabilities of occurrence, regardless of time period, location, and climate. A disadvan tage is that the SPI values may not be intuitive to decision makers. Also, equal categorical intervals have differing probabilities of occurrence. For instance, the probability differential between an SPI of -1.0 and -1.5 is 9.1% (moderate drought); between an SPI of -1.5 and -2.0, the probability differential is 4.4% (severe drought), even though both represent an index differential of 0.5.
Was this article helpful?