• SIMULATIONS: CanESM2-LE and ClimEx
• LANGUAGES: Python, Julia
• CONTACT: yann.chavaillaz@gmail.com
• DATE: 16th April 2019

1. Clustering of years in the CanESM2-LE
2. Correspondence between CanESM2-LE and ClimEx simulations
3. Merging of the ClimEx files for one simulation and one season
4. Removal of the climate change trend in ClimEx simulations
5. Computation of seasonal and extreme indicators
6. Sorting of years into clusters
7. Mask of specific regions
8. Plotting of figures

• Intermediate and final data, plus figures can be found on Neree in /exec/yanncha/sea_ice/.
• Functions are defined in files ./functions_***.py
• Climate variables: tasmin, tasmax, pr, prsn

• code: ./clustering/*
• method: K-means clustering
• input data: sea ice extent (sie - %) with a logarithmic transformation
• 3 clusters: no ice, ice, unclear
• The temporal distribution of clusters can be represented in a histogram by the code ./histogram_clusters.py
• The preparation of raw sie data to do the clustering is done with ./prepare_sic_september_forML.py

• file of correspondance: /exec/yanncha/sea_ice/clusters/sic_september_clusters_kmeans_CanESM2-LE.npy

• code: ./merge_months_[variable].py
• need to specify: months and season (SON, DJFM, AMJ)
• especially for pr: only hourly data is available. So first, transform hourly data to daily data with ./pr_hourlytodaily.py

• code: ./detrend_[variable].py
• need to specify: season
• method: cubic polynomial fit

• code: ./indices_[variable].py and ./residuals_[variable].py
• need to specify: season
• indicators for tasmin: minimum temperature of the season, 1st percentile, 5th percentile, seasonal average and season standard deviation
• indicators for tasmax: maximum temperature of the season, 99th percentile, 95th percentile, seasonal average and season standard deviation
• indicators for pr and prsn: daily maximum precipitation, 5-day maximum precipitation, seasonal average, seasonal standard deviation, relative -sum of precipitation over the season

• code: ./sort_indices.py
• need to specify: clim_var, indice, season
• use of the correspondence of clusters between CanESM2-LE and ClimEx of step 2

• code: ./mask_adminQC.jl
• shapefile: ./masking/admin_QC.[extension]
• need to specify: regions_name, regions_code, regions_number

• code: ./fig_deltamap_minmax.py
• need to specify: var, indice, indice_name, extrema (min, max, mean of the distribution) and units (of the indice)

• code: ./fig_distribution_qqplot.py
• need to specify: var, indice, indice_name and units (of the indice)

• code: ./fig_distribution_qqplot_regional.py
• need to specify: var, indice, indice_name, units (of the indice) and scale (specific region defined in step 7)

Steps 6, 7 and 8 for prsn.