Global NWP Skill - Origin of Past Improvements and Priorities For The Future

Global NWP skill has improved at a remarkably constant rate over the past few decades. Sources of major steps forward in skill are diverse, including: new observations (e.g. satellite radiances, GPSRO, aircraft, etc); improved use of existing observations (e.g. QC, bias correction, data assimilation); model improvements (e.g. physics, dynamical core) and changes made possible only via significant investment in high performance computing (e.g. resolution increases, ensemble prediction). This talk will begin with a review of the relative contributions of each area to improving Met Office global NWP skill over recent decades. Bringing things up to date, the relative impact of each observation type in the Met Office operational global NWP system since a major upgrade in March 2016 will also be reviewed.

The interpretation of improvements in global NWP skill depends crucially on the parameter, metrics and application of interest. Judged against traditional parameters e.g. 500hPa height, PMSL, global NWP might be viewed as a subject of diminishing returns in recent years. However, this masks the significant improvements associated with broadening of applications of global NWP over the past decade e.g. direct prediction of weather (e.g. cloud/precipitation analysis and prediction) and providing the boundary conditions for NWP at km-scale. Therefore, the true value of global NWP has to be assessed in this wider context. This talk will provide an assessment of global NWP skill improvements in this wider context.

Looking forward, the need for continued prioritization to maintain global NWP skill improvements is as important as it ever was and so a 'rolling review of requirements' process is essential. Where is the 'Achilles Heel' of the global NWP system? Which current/new applications or regional foci of global NWP provide the greatest potential return on investment? Is the solution new observations (if so, which, where and what?) or even greater investment in NWP to reduce biases and/or provide more diverse output? The answer is almost certainly both, especially given that global NWP is often only one of several applications of the global observing network (e.g. direct use by forecasters, etc). This talk will provide a summary of which areas the Met Office sees as priority areas for investment to improve global NWP in the next few years.