FairSprErr: Fair Spread to Error Ratio

In easyVerification: Ensemble Forecast Verification for Large Data Sets

Fair Spread to Error Ratio

Description

Compute the spread to error ratio (SPR) for probabilistic forecasts - not unlike the functions in SpecsVerification. SPR > 1 indicates overdispersion (underconfidence), whereas SPR < 1 indicates overconfidence in the forecasts.

Usage

FairSprErr(ens, obs)

Arguments

ens	n x k matrix of n forecasts for k ensemble members
obs	vector with n verifying observations

Details

Here we define the spread-error rate as the square root of the ratio of mean ensemble variance to the mean squared error of the ensemble mean with the verifying observations. We inflate the intra ensemble sample variance to account for the finite ensemble size as in Weigel (2011).

References

Weigel, A.P. (2012). Ensemble forecasts. Forecast Verification: A Practitioner's Guide in Atmospheric Science, Second Edition, 141-166.

See Also

veriApply, FairSprErr

Examples

tm <- toymodel()
FairSprErr(tm$fcst, tm$obs)

## compute spread to error ratio using veriApply
veriApply("FairSprErr", fcst = tm$fcst, obs = tm$obs)

## compare with 'unfair' spread to error ratio
veriApply("EnsSprErr", fcst = tm$fcst, obs = tm$obs)

easyVerification documentation built on Aug. 15, 2023, 9:06 a.m.