On compound probabilistic metrics with given minimal metrics

Finally, in order to compare the QTM results with models generated by other methods we have developed probabilistic metrics that formalize the notion of semantic coherence using probabilistic concepts and can be used to validate overlapping and incomplete clustering using multi-labeled corpora. They show that the proposed method can produce models of comparable, or even superior quality, than those produced with state of the art probabilistic methods.

Download Full-text

ENSO Prediction in Project Minerva: Sensitivity to Atmospheric Horizontal Resolution and Ensemble Size

Journal of Climate ◽

10.1175/jcli-d-14-00302.1 ◽

2015 ◽

Vol 28 (5) ◽

pp. 2080-2095 ◽

Cited By ~ 25

Author(s):

Jieshun Zhu ◽

Bohua Huang ◽

Ben Cash ◽

James L. Kinter ◽

Julia Manganello ◽

...

Keyword(s):

Climate Models ◽

Southern Oscillation ◽

Horizontal Resolution ◽

Prediction Skill ◽

Ensemble Size ◽

Enso Prediction ◽

Significant Difference ◽

Prediction Reliability ◽

The Impact ◽

Probabilistic Metrics

Abstract This study examines El Niño–Southern Oscillation (ENSO) prediction in Project Minerva, a recent collaboration between the Center for Ocean–Land–Atmosphere Studies (COLA) and the European Centre for Medium-Range Weather Forecasts (ECMWF). The focus is primarily on the impact of the atmospheric horizontal resolution on ENSO prediction, but the effect from different ensemble sizes is also discussed. Particularly, three sets of 7-month hindcasts performed with ECMWF prediction system are compared, starting from 1 May (1 November) during 1982–2011 (1982–2010): spectral T319 atmospheric resolution with 15 ensembles, spectral T639 with 15 ensembles, and spectral T319 with 51 ensembles. The analysis herein shows that simply increasing either ensemble size from 15 to 51 or atmospheric horizontal resolution from T319 to T639 does not necessarily lead to major improvement in the ENSO prediction skill with current climate models. For deterministic prediction skill metrics, the three sets of predictions do not produce a significant difference in either anomaly correlation or root-mean-square error (RMSE). For probabilistic metrics, the increased atmospheric horizontal resolution generates larger ensemble spread, and thus increases the ratio between the intraensemble spread and RMSE. However, there is little change in the categorical distributions of predicted SST anomalies, and consequently there is little difference among the three sets of hindcasts in terms of probabilistic metrics or prediction reliability.

Download Full-text