...
| Page Properties | ||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||||||||||||
|
| Expand | |||
|---|---|---|---|
| |||
|
...
Transport of ocean BGC tracers is ~70% of the cost of MPAS-O (when BGC tracers are turned on). We hypothesize that the advection step of the BGC tracer transport can be increased to 2-3 times that of the baroclinic time step with little loss in accuracy, thereby decreasing the overall cost of transport for BGC tracers in the ocean. We propose to implement a tracer-consistent, mass-conserving supercycling of the advection of all passive tracers in MPAS-O. If possible, options to (i) supercycle the tracer advection at time-steps larger than the sea-ice coupling interval (referred to as supercycling-over-coupling [SOC]) and (ii) subcycle the BGC reactions will be included. We propose to validate the scheme in MPAS by testing in a G-case configuration (ocean + ice), comparing mass conservation of carbon within the ocean, and comparing monthly averages of BGC tracer outputs.
Requirements
Requirement: Implement supercycling up to ocean-ice coupling time interval
Date last modified: 08/16/2021
Contributors: Andrew Bradley
Requirement: Implement supercycling of tracer over the ocean-ice coupling time interval (for use in spin-up of low-res)
Date last modified: 08/16/2021
Contributors: Andrew Bradley
Requirement: Implement nightly test in E3SM
Date last modified: 08/
1316/2021
Contributors: Katherine Smith
and Andrew BradleyAlgorithmic Formulations
Date last modified: 08/13/2021
Contributors: Andrew Bradley
Design and Implementation
Supercycling is implemented by accumulating the total mass flux when integrating the total density rho, using the b coefficients in the RK4 table combined with averaging over multiple steps. This mass flux is then used as the total mass flux when integrating mixing ratios q.
Date last modified: 08/13/2021
Contributors: Andrew Bradley
Planned Verification and Unit Testing
Verification and Unit Testing: Test tracer-consistency
Date last modified: 08/13/2021
Contributors: Andrew Bradley
Tracer consistency is tested by setting one debugTracer to a constant value and measuring its departure from that value in each cell in each time step.
Verification and Unit Testing: Test mass-conservation
Date last modified: 08/13/2021
Contributors: Andrew Bradley
Mass conservation is tested by measuring global mass at each time step and comparing with the first time step.
Planned Validation Testing
Validation Testing: Validating oceanic carbon mass-conservation
Date last modified: 08/13/2021
Contributors: Katherine Smith
Global simulations will be performed and comparisons of oceanic carbon mass-conservation will be compared between non-supercycled, suercycled, and supercycled over the ocean-ice coupling time interval will be compared. We expect mass-conservation to be comparable between all three.
Validation Testing: Validate against satellite products for ocean surface Chl-a/NPP
Date last modified: 08/13/2021
Contributors: Katherine Smith
and Andrew BradleyGlobal simulations will be performed and comparisons of ocean surface CHl-a will be compared between non-supercycled, suercycled, and supercycled over the ocean-ice coupling time interval will be compared. We expect minimal changes to occur when switching on supercycling (both under and over the ocean-ice coupling time interval).
Planned Performance Testing
Performance Testing: Determine performance by running MPAS-O with Supercycling turned on and off.
Date last modified: 08/13/2021
Contributors: Andrew Bradley and Katherine Smith
We will test the performance of the supercycling feature by running a
global simulation with and without supercycling turned on and compare timers of overall run time between the two. We expect a speed up of the run time with supercycling turned on.