hadjt
commented
8 months ago AMM15 appears to have much warmer bottom temperatures then other regional shelf sea models.
NOOS runs a multi model ensemble of operational northwest shelf models and AMM15 is much warmer under stratification. This is only an model intercomparison, and includes no observational truth. AMM15 has a more complex data assimilation scheme and higher resolution than most other models so it is possible (if unlikely) that other models are systematically biased.
There are limited EN4 observations on the North West shelf under stratification for the last few years, however a recent comparison to ICES data corroborates the warm bias
GASCO, a Norwegian gas pipeline company have used our bottom temperature forecasts for a number of years, and always found AMM7 to be better than AMM15. AMM15 NBTs are warmer than AMM7.
In a recent series of trials to test different lateral boundary conditions it was found that the current Copernicus PSY4 model LBC's tended to give a stronger bottom temperature bias then the Met Office ORCA12 LBC's. While this could be variability, it did allow us to examine how the bias was forming, in this particular case. It had been suggested that AMM15 mixes too strongly, and so the onset of stratification is delayed allowing the bottom temperatures to warm with the surface for a few more days, and then remain warmer while capped beneath the stratification.
End O'Dea had suggested the surface fluxes could be important as could the choice of the turbulence closure scheme. Jeff Polton had suggested the coupled waves could be important, and Matt Martin had suggested the data assimilation could be important. I wondered if the salinity bias in the Skagerrak, and the incorrect flow/surface gradient it suggests, could also be important
We're running a series of year-long trials to explore the causes of the bottom temperature biases, running with different surface fluxes, an alternate turbulence code closure scheme, with and without waves, with and without data assimilation. These trials also feed into issue #2.
The AMM15 operational system has significant biases in the near-bed temperatures which need to be investigated.