Comparison of Forecast Ensemble Spread to a Weather Feature Based Confidence Index

Forecast ensembles present an effective means of assessing the spread of forecast outcomes and thus the fragility or risk for error in a given forecast period.  However, ensembles require heavy computational resources and cost to create sufficient members to project a reasonable forecast variability spectrum. 

In response to this we have developed a Confidence Index (CI) which is an algorithm applied to a single forecast to determine risk of forecast error.  It is computationally less expensive compared to forecast ensembles.  When applied to a region CI detects the presence of weather features (“hazards”) that can lead to forecast error and creates a score that can be associated with forecast risk based on past forecast performance.  This technique is also applicable to other Hazard vs. Fragility systems.

Here we present a comparison of CI performance to forecast ensemble spread both to assess this alternative method in relation to ensembles and to determine that ensemble spread could be integrated into CI as another “hazard” that integrated into the CI algorithm.