GFDL - Geophysical Fluid Dynamics Laboratory

Improved Seasonal Prediction of Temperature and Precipitation over Land in a High-resolution GFDL Climate Model

March 6th, 2015

Liwei Jia, X. Yang, G.A. Vecchi, R.G. Gudgel, T.L. Delworth, A. Rosati, W.F. Stern, A.T. Wittenberg, L. Krishnamurthy, S. Zhang, R. Msadek, S. Kapnick, S. Underwood, F. Zeng, W. G. Anderson, V. Balaji and K. Dixon. Journal of Climate. DOI: 10.1175/JCLI-D-14-00112.1.

Summary

Skillful seasonal predictions of surface temperature and precipitation over land are in demand, due to their importance to ecosystems and sectors such as agriculture, energy, transportation. This study demonstrates skillful seasonal prediction of near-surface air temperature and precipitation over land using a new high-resolution climate model developed at GFDL, called FLOR. The study also diagnoses the sources of the prediction skill.

This study shows that the use of refined statistical analysis and a high-resolution dynamical model leads to significant skill in seasonal predictions of near-surface air temperature and precipitation over land. Output from predictions with FLOR is being made available to NOAA’s National Weather Service (and the world) through the North American Multi-Model Ensemble for Seasonal Prediction (NMME).

The authors employ a statistical optimization approach to identify the most predictable components of seasonal temperature and precipitation over land. The two most predictable components of near-surface air temperature are characterized by a spatially homogeneous component that is mostly due to changes in external radiative forcing in both boreal winter and summer, and a spatially heterogeneous ENSO-related pattern in boreal winter. The most predictable components of precipitation in boreal winter and summer are also ENSO-related. These predictable components of temperature and precipitation show significant correlation skill for all leads from 0 to 9 months. Importantly, the reconstructed predictions based only on the leading few predictable components from the model show considerably better skill relative to observations than raw model predictions.

 Figure 1: High-resolution GFDL-FLOR (middle panels) improves representation of observed (upper panels) connections between land precipitation and the El Nino-Southern Oscillation (ENSO) phenomenon, compared to its predecessor model (CM2.1, lower panels). Figure 2: Prediction skill for seasonal precipitation anomalies tied to ENSO, measured using SESS -- with which a perfect prediction has a value of 1 and values greater than zero indicate skill. The new high-resolution GFDL-FLOR (upper panel) is generally higher than for its predecessor model (CM2.1, lower panel).
Figure 1: High-resolution GFDL-FLOR (middle panels) improves representation of observed (upper panels) connections between land precipitation and the El Nino-Southern Oscillation (ENSO) phenomenon, compared to its predecessor model (CM2.1, lower panels). Figure 2: Prediction skill for seasonal precipitation anomalies tied to ENSO, measured using SESS — with which a perfect prediction has a value of 1 and values greater than zero indicate skill. The new high-resolution GFDL-FLOR (upper panel) is generally higher than for its predecessor model (CM2.1, lower panel).