Development of High-Resolution Japan Regional Reanalysis

For the purpose to detect climate change of mesoscale processes and realistically reproduce past extreme events over Japan, we are developing a regional reanalysis system covering Japan and its surrounding area. In this system, a nonhydrostatic model of the Japan Meteorological Agency (NHM-JMA) is one-way doubly nested (intermediate resolution model with 25km grid spacing and high resolution model with 5 km grid spacing) in the Japanese 55-year Reanalysis (JRA-55). A local ensemble transform Kalman filter is utilized to assimilate the observation data into both intermediate and high resolution models and it is modified to consider uncertainty in lateral boundary conditions and sparsity of observation data. The assimilated data are only the conventional observations available for recent 60 years, such as surface pressure and upper air observations, to keep the long-term consistency in reanalysis quality.

The assimilation system specially treats tropical cyclones (TCs). Assimilating only the surface observations and the upper air observations often failed to reproduce the TCs sufficiently over the ocean to the south and east of the Japan islands, because the observations are sparse. To cope with this difficulty, we introduce additional implementation of applying spectral nudging (SN) in the intermediate model to the outer JRA-55 and assimilating TC centre positions from the JMA’s best track data. The synergistic effects of additional applying the SN and assimilating TC centre positions significantly improve the reproducibility of TCs. The SN improves the flows steering TCs in the first guess fields, while the TC centre assimilation corrects the TC positions relative to the steering flows. The errors in TC centre positions are halved. The bias and threat scores of precipitation over the Japan islands are also improved from weak to heavy rains.

The regional reanalysis is an alternative to dynamic downscaling as a method to simulate mesoscale processes under control of large-scale atmospheric states. The result shows that the regional reanalysis is superior to dynamic downscaling in terms of reproducibility, although the former is much more expensive to the latter.

References

Fukui, S., et al., 2018: A Feasibility Study on the High-Resolution Regional Reanalysis over Japan Assimilating Only Conventional Observations as an Alternative to the Dynamical Downscaling. J. Meteor. Soc. Japan, 96, 565-585.