Observing-System Research and Ensemble Data Assimilation at JAMSTEC - Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications - page 500

observations taken during field campaigns are evaluated by data assimilation exper-
iments involving adding or removing observations. The results of these observing-
system experiments successfully demonstrate the value of the observations and are
highly useful for exploring the predictability of atmospheric disturbances.

21.1 Introduction

The Japan Agency for Marine-Earth Science and Technology (JAMSTEC) conducts
observations of the climate system over ocean and land in various parts of the
globe. JAMSTEC operates eight research vessels (Chikyu, Natsushima, Kaiyo,
Yokosuka, Mirai, Kairei, Hakoho-maru and Tansei-maru), Triangle Trans-Ocean
Buoy Network (TRITON) in the western tropical Pacific and eastern Indian Oceans,
Polar Ocean Profling System (POPS) in the Arctic Ocean and hundreds of Argo
floats. In addition, JAMSTEC conducts observations of the hydrological cycle of
the cryosphere and field campaigns to collect data on the atmosphere and ocean.
These various data sources provide a number of measurements daily, contributing
to the monitoring and investigation of the climate system.
JAMSTEC has one of the world's largest super computer systems devoted to the
earth sciences, the Earth Simulator. The Earth Simulator has been used to conduct
large-scale simulations in various areas in solid earth, ocean and atmospheric
sciences. The observational and computational capabilities at JAMSTEC provide
enormous opportunity for data assimilation. Ocean observations have already been
used to produce a number of data sets. A monthly global ocean analysis of Argo,
TRITON and CTD (conductivity, temperature and depth), called MOAA (Monthy
Objective Analysis using the Argo data) GPV (grid point value), was produced
by the optimal interpolation ( Hosoda et al. 2008 ). Argo was also used to con-
struct G-YoMaHa, with an objectively mapped velocity at 1,000 dvar ( Katsumata
and Yoshinari 2010 ) and MILA (Mixed Layer data set of Argo) GPV ( Hosoda
et al. 2010 ). The four-dimensional variational algorithm has been successfully
applied to an ocean general circulation model ( Masuda et al. 2003 ) and a global
coupled atmosphere-ocean model ( Sugiura et al. 2009 ). Atmospheric observations,
however, have not been fully utilized, indicating the potentials for data assimilation
and observing-system research.
A collaborative project was conducted from fiscal year (starting in April)
2006 to 2008 to develop an ensemble data assimilation system among the Japan
Meteorological Agency (JMA), JAMSTEC and the Chiba Institute of Science. The
system, named ALEDAS (AFES-LETKF ensemble data assimilation system), is
composed of the atmosphere general circulation model (AGCM) for the Earth
Simulator (AFES) as a forecast model ( Numaguti et al. 1997 ; Ohfuchi et al. 2004 ;
Enomoto et al. 2008 ) and the local ensemble transform Kalman filter (LETKF) as
an assimilation algorithm ( Hunt et al. 2007 ; Miyoshi and Yamane 2007 ). Obser-
vations were prepared from those used for numerical weather prediction at JMA
except for satellite radiances. An experimental analysis data set, called ALERA