Tuesday, 14 January 2020: 9:00 AM
151A (Boston Convention and Exhibition Center)
Leon Hermanson, Met Office Hadley Centre, Exeter, United Kingdom; and D. M. Smith, N. Dunstone, and R. Eade
Multi-year prediction is considered to get its predictability from persistence and dynamics in the ocean, particularly the Atlantic Ocean. The Decadal Climate Prediction Project (DCPP) is a coordinated multi-model investigation into decadal climate prediction, predictability, and variability included in CMIP6. For this the Met Office has developed the Decadal Prediction System version 4 (DePreSys4). The main innovation compared to the previous version of DePreSys is a new initialization method. Each of the ten forecast ensemble members is initialized with a different ocean initial condition. The Met Office Statistical Re-Analysis (MOSORA) creates a gridded ocean temperature and salinity data set by optimally interpolating using global covariances. The ensemble of analyses differ in their initial model covariances used for the interpolation. The aim is for the ensemble to represent some of the uncertainty arising from unobserved regions and depths in the ocean.
The ocean analyses are nudged into the coupled HadGEM3 model at a relatively high resolution for decadal prediction (0.25 degree ocean and sea-ice, ~60km in the atmosphere) to create the initial conditions for ten-year forecasts.
We find that observations in the North Atlantic since the 1960s do not constrain the range of decadal variability or the 20 year trends of the Atlantic Meridional Overturning Circulation (AMOC) in the ensemble. Here we explore the origins of the differences in the nudged AMOC, which are largely due to differences in salinity. We also present the impacts these differences have on decadal predictions.
The accompanying figure shows the five year smoothed anomalous AMOC at 45N in assimilation integrations where ocean potential temperature and salinity is nudged to the ocean analyses. Temperature and winds in the atmosphere are nudged to the same re-analysis (ERA-40 / ERA Interim). The general shape is the same with a peak in the late 1990s, but some assimilations have a range of more than 1.5 Sv (assim02, assim06) and others less than 0.75 Sv (assim04, assim09). There is one outlier (assim07) that has a different trend 1965 - 1995 than all the other members of the ensemble.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner