This project implements a method for detecing spatio-temporal consistent bias patterns in climate simulation output. Paper introducing the method can be found on https://doi.org/10.5194/gmd-2019-107.

• analyze.py is the main file. The core analysis logic is implemented in this file.
• BiasInstance.py and BiasFamily.py defines data structure for bias instance and bias family.
• PlotHeatMap.py defines a function to output a figure for a bias family, like Figure 3 in the paper.

The project can be run with python 3.5 and you may need to install some python packages, including Basemap, matplotlib, mlpy and netCDF4.

After the running environment has been set up, you can run the project with "python analyze.py".

All parameters used in this tool can be set at Lines 19-22 in "analyze.py".

• deltaA defines the minimum number of grid points that must appear in a bias instance. deltaA takes a value in [0, number of grid points].
• deltaR determines the minumum overlapping ratio that two bias instances in a family need to satisfy. deltaR takes a value in [0,1].
• gammaR is the significance level used to filter nonsignificant residuals. Smaller gammaR means more significant bias. gammaR takes the value in [0,1].
• gammaD is a paramter used to ensure the similarity of two bias instances in a family. Larger gammaD means stricter similarity requirement among bias instances in a family. gammaD takes a value in [0,infinity).