This ‘paleo-climate model’ talk will present a selection of results showing the evolution of simulation skill through six generations of the NCAR (National Center for Atmospheric Research) atmosphere model (CCM3 to CAM6) and five generations of the NCAR coupled model (CCSM2 to CESM2) spanning more than 20 years. During this time the complexity of the models have increased dramatically in almost every respect. These increases in complexity aim to target missing or poorly represented processes and interactions in models. However, an improved simulation, where new representations are thought to play a substantial role, is frequently not guaranteed. Although advances compared to early model versions are undeniable, they are not monotonic in nature, and certain degradations often come at the expense of including these new processes, a result of the inevitable trade-offs that come with model development.
Precipitation is often a main target of model validation and the tropical double ITCZ is crucial to model performance. The NCAR model has certainly had problems in the past, but in atmosphere-only simulations this was a minor problem prior to CAM4, became significant through CAM5 and was mostly remedied in CAM6. This recent degradation was the result of a trade-off due to the improved variability, ENSO and regional precipitation characteristics with the inclusion of deep convection modifications in CAM4.
Cloud radiative fields are important components of the global energy budget used to balance a fully coupled system. There is a stark contrast between short-wave cloud forcing, which has seen a 50% reduction in error, and long wave cloud forcing, which has seen virtually no improvement. This is somewhat surprising given that these fields are often addressed together as part of the model development process.
Variability at timescales shorter than monthly have until relatively recently been a more secondary consideration. Since CAM3 tropical variability in particular has improved, but this has not been even when different models of variability are considered e.g., Kelvin Waves and the Madden Julian Oscillation (MJO). However, other improvements such as blocking statistics show more robust improvement over time.
Finally, we present a brief overview of the recent development process for CAM6, the need to continually consider the whole coupled system and its, at times, ad-hoc nature.
3:00 pm – Welcome and intro - Antonio Navarra – CMCC
3:05 pm – Presenter’s talk - Richard Neale – NCAR
3:35 pm – Q&A
3:50 pm – End of Webinar
Working language: English
HOW TO PARTICIPATE:
The seminar will be broadcasted via Go-to-Webinar.
Please, click on the following link for registration:
After registering, you will receive a confirmation email containing information about joining the webinar.