This repository contains all the code necessary for replication of Mallett et al. (2021).
Mallett, Robbie DC, et al. "Faster decline and higher variability in the sea ice thickness of the marginal Arctic seas when accounting for dynamic snow cover." The Cryosphere 15.5 (2021): 2429-2450.
- Download daily ice-type data, average to monthly and regrid
This data is downloaded at daily time resolution from the Copernicus Data Store using a python script. The downloader script is available in processors/Download_Ice_Type_Data.ipynb This produces one giant folder with loads of files. The next steps of the analysis are done using ice_type/cds_daily_to_monthly_and_regrid.ipynb Here the files are dispatched to an individual folder structure with a folder for each month of each year The daily files for each month are then stacked and averaged using the command line tool cdo, which is called from within the notebook. This is what the p=subprocess.Popen() call does. Spicy! Finally all the monthly data for each year are aggregated into annual netcdf files using xarray.
- Make monthly mW99 2002-2018
To make mW99 you first need to make W99. This is done using the quadratic coefficients from Warren et al., 1999 (J. Clim). These coefficients are given in Warren_climatology.xlsx, and can be projected onto coordinates of your choosing using W99/W99_Gridder.ipynb W99 is then saved as a pickle file (e.g. W99_361.p for the analysis in this paper) Once you have the gridded W99 data, all you need to make mW99 is to combine it with the ice type data. This is done in W99/mW99_maker.ipynb. Annual netcdf files of mW99 are made using xarray
- Download daily SnowModel-LG data and average to monthly
The SnowModel-LG data required to reproduce the findings and figures in this paper can be downloaded from https://doi.org/10.5067/27A0P5M6LZBI You'll find 4x ~7 GB netcdf files. The original analysis for this paper was done using raw .gdat & .ctl files. There may be some references and allusions to these in the repo, but all analysis has now been redone from the raw .nc files for ease of replication. Please note, the lon/lat EASE coordinates are not provided with the .nc files - these can be found in mask.nc within the repo. When you've downloaded the SnowModel files, the first task is to convert them to monthly averages using SM_daily_to_monthly.ipynb.
- Download monthly radar freeboard data from CCI & AWI
First get as much data from CCI as possible:
Envisat data: https://catalogue.ceda.ac.uk/uuid/f4c34f4f0f1d4d0da06d771f6972f180
CryoSat2 data: https://catalogue.ceda.ac.uk/uuid/ff79d140824f42dd92b204b4f1e9e7c2
Now "top up" the CryoSat-2 data from the AWI ftp site, as the CCI data doesn't go quite up to 2018:
I use (from the command line) ncftp ftp://ftp.awi.de/sea_ice/product/cryosat2/v2p3/
- Download NESOSIM data
Download the files from here (only has up to 2015, which is all that was considered in the paper): https://earth.gsfc.nasa.gov/cryo/data/nasa-eulerian-snow-sea-ice-model-nesosim Process these into monthly mean files with Nesosim_monthly_processor.ipynb
- Calculate thickness components from the snow data sets and the radar freeboard data
This is done with processors/Calculate_thickness_components.ipynb and a bunch of helper functions from custom_tools that call the relevant data This makes annual netcdf files with the SIT components (snow and RF) for each month.
Figure 1: The regions of the Arctic Ocean are visualised using Fig_mask_map.ipynb
Figure 2: The drift-tracks and sampling of the NP drifting stations in the different regions are plotted in processors/Analyse_drift_tracks.ipynb
Figures 3, 4 & 5 are made using SM_Climatology.ipynb
Figure 6 is made using Regional_W99_vs_SMv2.ipynb (search for the function snow_contribution_timeseries_limited)
Figure 7 is made using the above notebook. Search for the function plot_radar_snow_correlations
Figure 8 is made using the above notebook - search for the function abs_var_bar (part a) and rel_var_bar (part b).
Figure 9 As above: abs_var_summary and rel_var_summary
Figure 10 As above: search for the timeseries function.
Figure 11 Same as above. Search abs_trends_summary (part a) and rel_trends_summary (part b).
Figure S1: Value of the propagation factor used to convert radar freeboard to ice freeboard, plotted as a function of snow density. This is made using Fig_linear_propagation.ipynb.
Figure S2: The number of valid, 25×25 km radar-freeboard grid cells in each region for each month. Made using Regional_W99_vs_SMv2.ipynb.
Figure S3: Difference in snow depth in SnowModel-LG when driven by ERA5 and Merra2 reanalysis data. Made using two notebooks: Compare_SM_reanalysis_difference1/2.ipynb
Figures S4 & S5:: Basinwide trends in first year ice extent as a fraction of total extent and mW99 SWE fields from 2003-2018 are made using Ice_type_analysis_and_testing.ipynb
Figure S6 to S12: made using Regional_W99_vs_SMv2.ipynb
Figure S13 made using SM_Climatology.ipynb
Figure S14 to S19: made using Regional_W99_vs_SMv2.ipynb