Supplement for 'A hybrid dynamical approach for seasonal prediction of sea-level anomalies: a pilot study for Charleston, South Carolina'
(c) 2022 All Rights Reserved
Authors: Thomas Frederikse [1], Tong Lee [1], Ou Wang [1], Ben Kirtman [2], Emily Becker [2], Ben Hamlington [1], Daniel Limonadi [1], Duane Waliser [1]
[1] NASA Jet Propulsion Laboratory, California Institute of Technology, Pasadena, USA
[2] University of Miami Rosenstiel School of Marine and Atmospheric Science, Miami, FL, USA
Please cite 'A hybrid dynamical approach for seasonal prediction of sea-level anomalies: a pilot study for Charleston, South Carolina' when using these scripts.
Next to these scripts, a data supplement with the adjoint sensitivities, the predictions, and statistics is available on Zenodo with DOI 10.5281/zenodo.6909128
This supplement contains the following directories:
The Scripts
directory contains the Julia scripts used to compute the convolution and determine the predictive skill of the various projections. These scripts rely on some external datasets, such as the ECCO forcings ECCO and the CCSM4 forcings part of the North-American multi-model ensemble (NMME).
These scripts perfrom various pre-processing tasks.
prepare_obs.jl
Read and prepare the tide-gauge and altimetry observations.regrid_CCSM_nn.jl
Re-grid the CCSM4 forcing data onto the ECCO LLC90 grid using a simple nearest-neighboor approach.regrid_functions.jl
Various functions used byregrid_CCSM_nn.jl
.read_CCSM_ssh_projections.jl
Read the direct SSH predictions from CCSM4.
compute_convolution.jl
# Compute the convolution between various forcings and the adjoint sensitivities.create_persistence_projections.jl
# Create the damped persistence predictions.
postprocess_convolution.jl
# Post-process the convolution and other predictions and compute statistics (ACC, RMSE etc.).convert_sens_to_2d.jl
# Re-grid adjoint sensitivities from LLC90 to a 2-dimensional grid for plotting purposes
This directory contains the figures from the paper and supplement and the GMT scripts and input data to reproduce them. You can download and build GMT from here. All plots have been made using GMT version 6.2.