1. Introduction

This repository contains the code for the project "More extreme summertime North Atlantic Oscillation (NAO) under climate change".

2. System Requirements

The code has been tested on Linux system.

enviroment.yml provides a list of python dependences. an environment can be created by following code: bash conda env create -f environment.yml conda activate Tel_MMLE

3. Data

All data used in this study are publicly available. The large ensembles are available at the MMLE Archive. The NOAA-CIRES-DOE 20th Century Reanalysis (20CR) data are aviailable at the NOAA Physical Sciences Laboratory. In order to use more high-level class objects defined in this project (such as those defined in /work/mh0033/m300883/Tel_MMLE/src/MMLE_TEL/index_generator.py), the data should be put under data directory, and the directory structure should be like: ├── CanESM2 # first level is the name for the model / reanalysis ├── CESM1_CAM5 ├── CR20 ├── CR20_allens ├── ERA5 ├── ERA5_allens ├── GFDL_CM3 ├── MK36 ├── MPI_GE ├── MPI_GE_onepct │ ├── codes # second level contains the pre-processing codes \n │ ├── composite # some results such as composite analysis │ ├── EOF_result # EOF analysis results │ ├── zg # variables such as 'zg', 'ts' │ ├── zg_Aug # also variables that are separated by months │ ├── zg_Jul │ ├── zg_Jun │ ├── ts └── zodes_for_all # some pre-processing codes that apply to all models Note that the data need to be pre-processed before calculations. Some pre-process codes can be found in zodes_for_all. However, I recommend you to request an example data from me (the data size is too big for GitHub) so that the pre-process history can be found using cdo showattribute.

4. Code structure

The project is mainly written in python. Some of the pre-processing and post-processing are in bash.

The codes include script, src and test directories. The script directory contains the main scripts for the calculations and ploting, which import some of the functions and class from the src directory. The src contains the more basic level calculations. The code structure is as follows: ├── script │ ├── calculations_paper # main calculations for the paper │ │ ├── 1generate_index # generate the NAO index for LEs, reanalysis, and resample the MPI_GE (100) to smaller ensemble sizes. │ │ ├── 2extreme_count # count the extreme events for each model, reanalysis, and resampled ensemble size │ │ ├── 3composite_analysis # composite analysis of surface temperature during the extreme NAO events │ │ ├── 4project_index # project the geopotential height field onto the NAO spatial pattern for the projected index │ │ └── 5statistical_cal # some statistical calculations such as calculate slope. │ ├── plots_paper │ │ ├── main_paper.py # to generate the plots in main text of the paper │ │ └── supplementary_paper.py # to generate the plots in supplementary material of the paper │ └── _explore_script # some scripts that may be useful for exploring the data ├── src │ ├── composite # The major package for the composite analysis │ ├── compute │ ├── EVT │ ├── extreme # The major package for the extreme event counts │ ├── html │ ├── __init__.py │ ├── MMLE_TEL # the major package for NAO index generate and statistical analysis │ ├── obs │ ├── plots # source code for ploting │ ├── quick_plot │ ├── reanalysis │ ├── Teleconnection # source code for decomposing the NAO pattern │ └── warming_stage test directory contains test codes for some basic functions in src directory, with (simulated) mini dataset.

5. How to start

You can easily copy the code to your local machine / HPC by running the following command: bash git clone https://github.com/liuquan18/Tel_MMLE.git cd Tel_MMLE To repeat the calculation, I strongly recommend using the same conda environment as me. The environment file is provided in the repository, and an environment can be created by following code: bash conda env create -f environment.yml conda activate Tel_MMLE To make the codes in scr directory as importable modules, you need to run the following command: bash python3 -m pip install -e . Then you can run the scripts in the script directory to repeat the calculations and generate the plots.

6. Demo

The analysis follows "indexgenerate" --> "extremecount" --> "composite". For model = 'MPI_GE', a quick calculation could be:

1. For "indexgenerate": go to file `script/calculationspaper/1generateindex/1MMLEindexgenerator.py, a function calledindexgene(model)` can be used to generate index for one model (at specific plev).

2. For "extreme count": in file script/calculations_paper/2extreme_count/1MMLE_extreme_count.py, a function called extreme_count(model) can be used to count extreme events for one model (at specific plev).

3. For "composite analysis": in file script/calculations_paper/3composite_analysis/1MMLE_composite_analysis.py, a function called composite(model) can be used to do composite analysis.

The "indexgenerate" would take relatively long time, depending on the size of years and ensemble members that used (e.g, 10 years and 100 ensemble members for MPIGE would take more than half an hour). The "extreme count" can be finished instantly. "composite analysis" would also take very long time to finish.