# Calculation of the AFWA Performance Index

## Methodology

1. ### Calculation of skill scores

Once the forecast and persistence rms errors have been determined for a particular combination of parameter/forecast range and for a particular day, the daily skill score may be calculated. This is defined in terms of Reduction of Variance, i.e., where:

• rf is the rms forecast error
• rp is the rms persistence error

The smaller the ratio between forecast and persistence errors, the closer the skill score will be to 1 (perfection). If the forecast error is greater (worse) than the persistence error, then the skill score will be negative.

2. ### Compilation of the index

There are several further stages in the compilation of the Index from the skill scores. Firstly a weighted average, Sj, of the individual skill scores for day j is calculated, where:

• wi is the weight for the i-th component (see table below)
• SSi,j is the skill score for the i-th component for day j.

The following table defines the weights, wi, for each parameter/level/forecast hour. For models initialized at 0Z, the 12hr, 24hr, 36hr, and 48hr skill scores and weights are used. For models initialized at 6Z, the 6hr, 18hr, 30hr and 42hr skill scores and weights are used.

Weights for Each Parameter / Level / Forecast Hour
6/12 hr 18/24 hr 30/36 hr 42/48 hr
400WS 4 3 2 1
250WS 4 3 2 1
850WS 4 3 2 1
700TD 8 6 4 2
400TD 8 6 4 2
400TT 4 3 2 1
400HT 4 3 2 1
SLP 8 6 4 2
SFCTT 8 6 4 2
850TD 8 6 4 2
SFCTD 8 6 4 2
SFCWS 8 6 4 2

Once the weighted sum, Sj, has been calculated the daily index value, Nj (Go Index), is defined as:

Nj is the Go Index.

Modified January 6, 2011 8:01 PM
NOAA | NWS | NASA | Navy Weather
Air Force Weather | Website owner: JCSDA