Premium
Impact of altimetry data on ENSO ensemble initializations and predictions
Author(s) -
Zheng Fei,
Zhu Jiang,
Zhang RongHua
Publication year - 2007
Publication title -
geophysical research letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.007
H-Index - 273
eISSN - 1944-8007
pISSN - 0094-8276
DOI - 10.1029/2007gl030451
Subject(s) - data assimilation , hindcast , ensemble kalman filter , initialization , forecast skill , el niño southern oscillation , ensemble learning , altimeter , climatology , predictability , environmental science , ensemble forecasting , sea surface temperature , sea surface height , multivariate enso index , meteorology , kalman filter , computer science , mathematics , geology , statistics , la niña , extended kalman filter , machine learning , physics , programming language
The El Niño/Southern Oscillation (ENSO) predictions strongly depend on the accuracy and dynamical consistency of the coupled initial conditions. Based on the proposed ensemble Kalman filter (EnKF), a new initialization scheme for the ENSO ensemble prediction system (EPS) was designed and tested in an intermediate coupled model (ICM). The inclusion of this scheme in the ICM leads to substantial improvements in ENSO prediction skill via the successful assimilation of both observed sea surface temperature (SST) and TOPEX/Poseidon/Jason‐1 (T/P/J) altimeter data into the initial ensemble conditions. Comparisons with the original ensemble hindcast experiment show that the ensemble prediction skills were significantly improved out to a 12‐month lead time by improving sea level (SL) initial conditions for better parameterization of subsurface thermal effects. It is clearly demonstrated that improvement in forecast skill can result from the multivariate and multi‐observational ensemble data assimilation.