Skip to content

Latest commit

 

History

History
132 lines (85 loc) · 11.7 KB

README.md

File metadata and controls

132 lines (85 loc) · 11.7 KB
Raw

Travis-CI GHActions Documentation Status lcov

My personal changes have been made within the WIM branch, this branch implements the work from Bennetts et al. (2017) into CICE6.

The CICE Consortium sea-ice model

CICE is a computationally efficient model for simulating the growth, melting, and movement of polar sea ice. Designed as one component of coupled atmosphere-ocean-land-ice global climate models, today’s CICE model is the outcome of more than two decades of community collaboration in building a sea ice model suitable for multiple uses including process studies, operational forecasting, and climate simulation.

<<<<<<< Updated upstream

CICE6 includes the capabilities for measuring the impacts of waves on sea ice across a global scale. In initial testing of CICE6 it was apparent that there is no code for the propagation and attenuation of wind waves and swell through sea ice. Without waves no ice breakup will occur and all new floes are assumed to have the geometry of pancake ice, rather than using the theory of Shen et al. (2001). The Waves-in-Ice Module (WIM) module is a way to propagate waves through the ice pack in CICE6, such that preliminary runs can be completed using CICE's new FSD routine. This module is from the code used in Bennetts et al. (2017) (https://bitbucket.org/puotila/cicewithwaves/src/master/), although some modifications have been made to make it compatible with the latest version of CICE. It was decided that CICE's ice breakup routine (Roach et al., 2018; Horvat et al., 2015) would be used within this model to keep comparisons as simple as possible. The short-coming of this decision is that wave-sea ice interactions are not coupled, we are currently investigating the effects of this.

In our testing we found that the new floe subroutine (Shen theory) is highly influential on the floe size of the ice cover. As such, applying a north-south wave cut-off (using mean wave direction) was not sufficient. As the wave field in CICE has no memory, a large swarth of cells would frequently 'switch' between zero wave energy (waves travelling north) and non-zero wave energy (waves travelling south). This resulted in new floes being assigned to the largest floe size category in the first epoch, before being rapidly broken up by the wave field in the second epoch, causing the floe size diagnostics terms to fluctate. As such, we have applied a parameteric spreading technique paired to calculate the 2D directional spectrum, from which we intergrate over the wedge travelling southward to more accurate approximate the resulting wave energy passing into the ice cover.

Stashed changes

This repository contains the files and code needed to run the CICE sea ice numerical model starting with version 6. CICE is maintained by the CICE Consortium. Versions prior to v6 are found in the CICE-svn-trunk repository.

<<<<<<< Updated upstream CICE consists of a top level driver and dynamical core plus the Icepack column physics code], which is included in CICE as a Git submodule. Because Icepack is a submodule of CICE, Icepack and CICE development are handled independently with respect to the GitHub repositories even though development and testing may be done together.

The first point of contact with the CICE Consortium is the Consortium Community Forum. This forum is monitored by Consortium members and also opened to the whole community. Please do not use our issue tracker for general support questions.

  1. The ice edge is taken to be the first equatoward cell with less than puny SIC, although this threshold parameter can be changed.
  2. Read in data for these respective cells from a WW3 hindcast dataset (https://data.csiro.au/collections/collection/CI6616v010).
  3. Use a Bretscheider spectrum to define the wave spectrum, $S(\omega)$, for each cell along the edge of the wave mask (ice_floe.F90/increment_floe_long).
  4. Apply a parametric spreading technique (cosine-sqaured) to approximate a 2D wave-energy, $E(\omega,\theta) = S(\omega)D(\theta)$ (ice_floe.F90/increment_floe_long).
  5. Extract the component heading southward $E(\omega,\theta=\pi)$, (ice_floe.F90/increment_floe_long).
  6. Propagate waves given the significant wave height, peak period, and mean wave direction into the ice pack.
  7. Feed this data into the icepack thermodynamic routine.

Attenuation is computed as per the observation of Meylan et al. (2014) currently, there is the code in place to implement the attenuation model presented in Williams et al. (2013a,2013b), but this has not been tested within CICE6 yet.

For more details see https://noahday.notion.site/README-af9c753fb9ab47f2a0220bda3cfa256d.

Stashed changes

If you expect to make any changes to the code, we recommend that you first fork both the CICE and Icepack repositories. In order to incorporate your developments into the Consortium code it is imperative you follow the guidance for Pull Requests and requisite testing. Head over to our [Contributing][contributing] guide to learn more about how you can help improve CICE.

<<<<<<< Updated upstream [contributing]: https://github.com/CICE-Consortium/About-Us/wiki/Contributing

Useful links

Stashed changes

<<<<<<< Updated upstream

License

See our License and Distribution Policy.

Shen, H.H., Ackley, S.F., Hopkins, M.A., 2001. A conceptual model for pancake-ice formation in a wave field. Ann. Glaciol. 33, 361–367. https://doi.org/10.3189/172756401781818239

Williams, T.D. et al. (2013) ‘Wave–ice interactions in the marginal ice zone. Part 1: Theoretical foundations’, Ocean Modelling, 71, pp. 81–91. doi:10.1016/j.ocemod.2013.05.010.

Williams, T.D. et al. (2013) ‘Wave–ice interactions in the marginal ice zone. Part 2: Numerical implementation and sensitivity studies along 1D transects of the ocean surface’, Ocean Modelling, 71, pp. 92–101. doi:10.1016/j.ocemod.2013.05.011.

❓FAQ

  • What is the domain of the module?

    The WIM is currently limited to only the Southern Hemisphere as there are more complexities with land around the Arctic.

  • What dimension are the wave spectra?

    Currently, the WIM is limited to 1D (along longitudinal lines). There is some infrastructure for moving to a 2D wave propagation scheme, although Alberto Alberello and Luke Bennetts ran into problems with the transfer of wave energy between blocks in CICE.

  • What is the wave spectrum made up of?

    Using significant wave height and a peak period a Bretschneider spectrum is defined. This spectrum is then propagated either North or South according to it's mean wave direction.

  • Where are waves propagated from?

    Wave statistics are taken from the closest ice-free cell. It uses a 1D wave scattering advection scheme, so it reads in wave direction but theses directions don't evolve over time.

  • What output does this module produce?

    Wave spectrum, peak period, significant wave height for each cell.

  • What FSD do they use?

    They use the prognostic FSD routine in CICE6 (Roach et al., 2017). However it could be possible to implement a split power law to describe the FSD following Toyota et al. 2017 (as seen in Bennetts et al., 2017) for the breakup routine.

  • Does it use wave direction?

    They do use wave direction from the closest ice free cell. The wave spectrum is 'advected' through the ice. I say 'advected' as it is really only attenuation rather than full advection. And the wave direction does not change as the waves collide with floes and other waves.

  • Relevant variable dimensions

    real (kind=dbl_kind), dimension (nx_block,ny_block,max_blocks) :: &
         swh     , & ! significant wave height (m)
         mwd     , & ! mean wave direction (Rads)
         ppd         ! wave peak period (s)

  • What are the attenuation settings?

    There are two kinds of attenuation present in this module, the simplest is based off of observations from Meylan et al. (2014). The more complex (floe size dependent) attenuation model has not currently been tested in CICE6 and is based off of the work by Williams et al. (2013a,2013b).

Extra techinical details

increment_floe is the main script in the module, it calls all other files in the wave_ice_code and exists within CICE code. This code first initalises: frequency min/max, angular frequencies, frequency, wave direction, initial energy spec, wavelength, wave number, dummy spectral moments, wave spectrum tolerance, spectrum passed through ice-covered ocean, sine mean wave direction for rows, values required for calculations of mean wave direction, WIM termination flags, attenuation parameters, and points in frequency and angular domains.

This code proceeds by defining the wave mask edge and propagate waves into ice, define attenuation coefficient coefficients. Then it calculates the wavelengths and wave number by lambda = gT^2/2pi, kappa = 2pi/lambda as well as direction th_in = -pi/2 + (lp_i-1)pi/(nth_in-1). Then the Bretschneider SDF function is called to initialise the wave spectrum.

This gets the code to the point where it can update the wave spectrum and the floe sizes. This can work in both the coupled and uncoupled set ups by calling sub_Balance or sub_Uncoupled respectively. sub_Balance first calculates the mass,and fr, where mass is the scaled mass and fr is the flexural ridging (note that dom and dth are d*omega and d(th)). It then implements Simpson's rule for dth and calculates lambda_ice. Its advection scheme differs from sub_Uncoupled as it uses the functions fn_IntAttn and fn_AvAttn to calculate the integrated attenuation coefficient with respect to floe diameter and the averaged attenuation coefficient with respect to floe diameter respectively. This calculation takes the maximum floe size, lambda_ice(lp_i) and om(lp_i) as input. The wave attenuation is handled in the same exponential (+ cosine) scheme as in the previous subroutine (although with different coefficient values).

Stashed changes