A summary of the GPCP precipitation-based ENSO indices (EI, LI, ESPI)

The goal in constructing a precipitation-based measure of ENSO was to estimate the gradient of rainfall anomalies across the Pacific basin and ensure a good relationship with SST- and pressure-based indices. Areas were selected that represent the Maritime Continent (MC) (10° N – 10°S; 90° E – 150° E) and central to eastern Pacific (P) (10° N – 10° S; 160° E – 100° W). These regions capture the largest precipitation anomalies associated with the interannual variations of the Walker circulation and contain the largest correlations between GPCP and Nino 3.4 and SOI. Within P and MC the absolute magnitude of the largest correlation is over +0.6.

Because of the spatially varying nature of rainfall, it was decided to use a moving block average which would capture the strongest zonal gradients within the equatorial Pacific. This procedure is unlike many fixed area average indices, and allows for a realistic meridional component of the precipitation gradient and migration of the ascending and descending branches of the Walker circulation.

A 10° latitude by 50° longitude box is moved in 2.5° increments (grid block by grid block) throughout the P and MC domains. The maximum and minimum average precipitation anomalies are found for P (Ap+ and Ap- respectively) and MC (Amc+ and Amc- respectively). To create a homogeneous record, Ap+, Ap-, Amc+, and Amc- were separately averaged over 1991-97 for GPCP and GPROF. Then, the GPROF values were subtracted from the GPCP values to create adjustments that were applied to the GPROF part of the record (usually the last few months). Significant adjustments are made to Ap+, and especially in the winter and spring months. However, it is planned to replace GPROF values with the more accurate GPCP analyses as they become available, about three months after the observation time. Amc- is subtracted from Ap+ and normalized to create the El Niño Index (EI). Ap- is subtracted from Amc+ and normalized to create the La Niña Index (LI). Normalization is achieved for each month by subtracting off the 1979-98 means and dividing by the standard deviations as shown in Eq. 1:

(1)

where i = (January, February, … , December). Positive EI (LI) values would indicate that the ENSO cycle was in its warm (cold) phase. There are advantages to quantifying the evolution of the warm and cold phases of ENSO through separate indices. However, there is also an advantage to having one index describe the ENSO cycle. This was accomplished by taking the difference and normalizing to create the ENSO Precipitation Index (ESPI):

ESPI = normalized EI – LI (2)

 Positive (negative) values indicate the warm (cold) phase of the ENSO cycle. Applying a two month running mean to the ESPI data reduces the effect of the 30-60 day oscillation signal.

REFERENCE:

Curtis, S. and R. Adler, 2000: ENSO indexes based on patterns of satellite-derived precipitation. J. Climate, 13, 2786-2793.