Which similarity metrics are the most helpful to understand climate?
Randomized Dependence Coefficient - Contains the implementation of the 'randomized dependence coefficient' similarity measure
Similarity Measures - Contains a summary of all the time series similarity measures found in literature
Writing in Science - Contains summaries for the 'Writing in Science'-Course on Coursera
data - Contains the data used in this work
papers - Contains summaries of related work
0_vizualize.ipynb - Visualizes the data using the matplotlib basemap toolkit
10_agreeableness_defined_with_std (Widget version).ipynb - Puts at disposal a widget to try different value- and agreement-threshold combinations
10_agreeableness_defined_with_std_and_entropy.ipynb - Presents the notion of agreeableness between similarity measures
11_level_of_agreement.ipynb - Presents the initial idea with level of agreement
12_time_delayed_dependencies.ipynb - Computes and plots the dependencies delayed in time
13_dependencies_to_different_levels.ipynb - Computes and plots the dependencies over different pressure levels
14_time_delayed_dependencies_and_agreeableness_to_different_levels.ipynb - Computes and plots the dependencies and the agreeableness between different similarity measures over levels and delayed in time
15_applications.ipynb - Shows some further applications for agreeableness
1_plotting_1979-2019.ipynb - Plots the mean per longitude of the u wind component for every latitude of every year
2_point-wise_similarities.ipynb - Computes the similarity to a random point for every point on the map
3_deriving_QBO.ipynb - Derives different QBO indices from the data
4_converting_coordinates_to grid_indices.ipynb - Converts geographical coordinates into the corresponding indices for the gaussian grid
5_similarity_to_qbo.ipynb - Computes the Mutual Information and Pearson Correlation between the QBO and the whole map at 30 hPa (for the whole period, for the whole period but every month seperately and for the whole period but only winter months are taken into account)
6_different_similarity_measures.ipynb - Presents the different similarity measures implemented in the similarity_measures.py file
7_compare_similarity_measures.ipynb - Presents different functions to make similarity measures comparable
8_combine_similarity_measures.ipynb - Presents different functions to combine the similarity measures
9_dependencies_between_similarity_measures.ipynb - Visualizes the relationship between different similarity measures
bachelor_thesis_figures.ipynb - Contains all the figures used in the thesis
calculations.py - Is the module for the similarity and agreeableness computation between time series similarity measures.
combining.py - Is the module containing different functions to combine results of similarity measures
comparing.py - Is the module containing different functions to make results of similarity measures comparable
environment.yaml - Contains an anaconda environment with all the required packages
plots.py - Is the module for visualizations of the similarity and agreeableness computation results.
proposal_figures.ipynb - Contains the figures used in the proposal presentation
similarity_measures.py - Is the module containing different similarity measures for time series
Create a modular framework for analysing and understanding relationships in climate data. The moduls should allow for testing different climate indices, similarity metrics, and time-scales. The results should be interpretable by climate scientists.
The the u component (east-west) of the wind for 512 longitudes, 256 latitudes, and 37 altitudes aggregated by months.
era-int_pl_1981-mm-u.nc
file from this repository.era-int_pl_1981-mm-u.nc
in the directory data/
.era-int_pl_1979-2019-mm-u.nc
file from this repository for the "Data Preprocessing" stepUse the data manipulation tool cdo to extract all values for pressure level 3, 30, 70 and 300 hPa. Example for extracting values for pressure level 30 with the following command:
cdo -select,level=30 era-int_pl_1979-2019-mm-u.nc era-int_pl_1979-2019-mm-l30-u.nc
Analogously for 3, 70 and 300
conda env create -f environment.yaml
conda activate climate_similarity_measures
Randomized Dependence Coefficient
folder and run python setup.py install
Values at Singapore (1N, 104E) at pressure level 30 hPa
Values at +- 5°north and south of the equator at pressure level 30 hPa
Data to cross check can be found under the following link:
https://acd-ext.gsfc.nasa.gov/Data_services/met/qbo/QBO_Singapore_Uvals_GSFC.txt
The file has to be placed in the directory data/
You have to remove the first 9 lines in order to make it readable for Python.
If you want to skip this step, you can download a preprocessed version here