A new verification technique has been developed to more fairly evaluate forecasts with spatial or temporal errors. This method differs from traditional techniques in the way it associates forecasts and observations. It forms each forecast-observation pair by minimizing a cost function calculated between a target datum and a field of candidate data.

In this scheme, forecast skill is measured by two metrics called acuity and fidelity. Acuity and fidelity differ only in the roles taken by the observations and forecasts during the calculation of the cost function. Acuity is the value of the cost function when the observations take the target role and the forecasts take the role of the candidate field. Fidelity is the value of the cost function when these roles are reversed. Qualitatively, acuity represents the skill of the forecast at predicting the features of the observed data, while fidelity represents the faithfulness of the forecast's predictions to the observed data.

In this study, the acuity-fidelity technique is applied to the verification of one-hour accumulated precipitation forecasts from several numerical weather prediction models. A sensitivity study of the scheme's configurable parameters is performed and reasonable values are determined.

It is demonstrated that acuity-fidelity measures the skill of precipitation forecasts in a way that is consistent with subjective verification based on visual inspection, particularly if the metrics are stratified by intensity. Visualization of the components of acuity and fidelity can be used as a tool for exploring and characterizing the skill of a forecast.

Acuity-fidelity metrics are compared to threat score statistics for a case study of approximately one month of precipitating weather events. Acuity-fidelity is shown to provide a more fair and meaningful assessment of the forecasts, and may provide a way to objectively assess forecasts that previously have been amenable only to subjective verification.