# -*- coding: utf-8 -*- """ Taylor Diagram =================================== The Taylor Diagram is a graphical tool introduced by American meteorologist Karl E. Taylor in 2001 to assess the performance of models. It provides an intuitive way to compare how well model simulations match observed data and to compare the performance of different models. The basic structure of the Taylor Diagram includes radial and tangential coordinates. In the chart, the standard deviation of observed data is placed at the center, while the standard deviation of model simulations extends radially outward. Additionally, the correlation coefficient, which measures the strength and direction of the linear relationship between two sets of data, is represented along the tangential axis. This chart allows for a visual comparison of the dispersion, shape, and correlation with observed data. The advantages of the Taylor Diagram in research, particularly in comparing climate model simulation performance, are as follows: - **Intuitiveness**: The Taylor Diagram presents the similarity between model simulations and observed data in an intuitive way, making it easy for researchers, including non-experts, to quickly understand model performance. - **Comprehensive Comparison**: The Taylor Diagram allows for the simultaneous comparison of model simulations with observed data and the performance of multiple models. This aids in identifying the most accurate models. - **Comprehensive Assessment**: By combining key statistical measures such as standard deviation and correlation coefficient, the Taylor Diagram offers a comprehensive evaluation of model performance, considering multiple aspects rather than focusing on a single parameter. - **Wide Applicability**: The Taylor Diagram is not limited to climate model performance; it can be applied to evaluate models in various fields such as meteorology, oceanography, and environmental science. The Taylor Diagram, as a graphical tool, provides a clear and comprehensive way for researchers, including those without a specialized background, to assess model performance. It is particularly useful for comparing multiple models and complex datasets in research. .. seealso:: Taylor, K. E. (2001), Summarizing multiple aspects of model performance in a single diagram, J. Geophys. Res., 106(D7), 7183–7192, doi: https://doi.org/10.1029/2000JD900719. Before proceeding with all the steps, first import some necessary libraries and packages """ import easyclimate as ecl import xarray as xr import numpy as np import matplotlib.pyplot as plt import pandas as pd # %% # Assuming the simulation results of model 'a' are as follows da_a = xr.DataArray( np.array([[1, 2, 3], [0.1, 0.2, 0.3], [3.2, 0.6, 1.8]]), dims=("lat", "time"), coords={ "lat": np.array([-30, 0, 30]), "time": pd.date_range("2000-01-01", freq="D", periods=3), }, ) da_a # %% # At the same time, we also assume that model 'b' has the following simulation results da_b = xr.DataArray( np.array([[0.2, 0.4, 0.6], [15, 10, 5], [3.2, 0.6, 1.8]]), dims=("lat", "time"), coords={ "lat": np.array([-30, 0, 30]), "time": pd.date_range("2000-01-01", freq="D", periods=3), }, ) da_b # %% # Observational (real) data should be directly obtained from instruments or reanalyzed in real life. # Here we simply set it as the linear relationship between model `a` and model `b`. da_obs = (da_a + da_b) / 1.85 da_obs # %% # Build Dataset # ------------------------------------ # :py:func:`easyclimate.plot.calc_TaylorDiagrams_metadata ` provides us # with the necessary parameters for calculating the subsequent Taylor diagram. taylordiagrams_metadata = ecl.plot.calc_TaylorDiagrams_metadata( f=[da_a, da_b], r=[da_obs, da_obs], models_name=["f1", "f2"], weighted=True, normalized=True, ) print(taylordiagrams_metadata) # %% # Basic Figure Framework # ------------------------------------ # :py:func:`easyclimate.plot.draw_TaylorDiagrams_base ` can # draw the basic framework of the Taylor diagram, which provides a basic drawing board for the data we will place. fig, ax = plt.subplots(subplot_kw={"projection": "polar"}) ecl.plot.draw_TaylorDiagrams_base(ax=ax, std_max=2.5) # %% # Dataset Points # ------------------------------------ # Try using :py:func:`easyclimate.plot.draw_TaylorDiagrams_metadata ` to place # data on the basic framework of the Taylor diagram. # # .. note:: # :py:func:`ax.legend() ` or :py:func:`plt.legend() ` can add legend for Taylor diagram. # fig, ax = plt.subplots(subplot_kw={"projection": "polar"}) ecl.plot.draw_TaylorDiagrams_base(ax=ax, std_max=2.5) ecl.plot.draw_TaylorDiagrams_metadata( taylordiagrams_metadata, ax=ax, marker_list=["o", "+", "*"], color_list=["black", "red", "green"], label_list=["", "", ""], legend_list=taylordiagrams_metadata["item"].to_list(), cc="cc", std="std", ) ax.legend(bbox_to_anchor=(1, 0.9)) # %% # The parameter `half_circle = True` in :py:func:`easyclimate.plot.draw_TaylorDiagrams_base ` can make the entire Taylor drawing board # appear in a semi circular state, which allows us to discover # data points with negative standardized standard deviation (marked with a red cross) fig, ax = plt.subplots(subplot_kw={"projection": "polar"}) ecl.plot.draw_TaylorDiagrams_base( ax=ax, std_max=2.6, std_interval=0.5, half_circle=True, x_label_pad=0.5, arc_label_pad=0.5, ) ecl.plot.draw_TaylorDiagrams_metadata( taylordiagrams_metadata, ax=ax, marker_list=["o", "+", "*"], color_list=["black", "red", "green"], label_list=["", "", ""], legend_list=taylordiagrams_metadata["item"].to_list(), cc="cc", std="std", ) # %% # Points' Labels # ------------------------------------ # `label_list` in :py:func:`easyclimate.plot.draw_TaylorDiagrams_metadata ` can # specify the labels of these data points fig, ax = plt.subplots(subplot_kw={"projection": "polar"}) ecl.plot.draw_TaylorDiagrams_base(ax=ax, std_max=2.5) ecl.plot.draw_TaylorDiagrams_metadata( taylordiagrams_metadata, ax=ax, marker_list=["o", "+", "*"], color_list=["black", "red", "green"], label_list=["1", "", "3"], legend_list=taylordiagrams_metadata["item"].to_list(), cc="cc", std="std", ) # %% # The position of these labels can be finely adjusted using `point_label_yoffset` and `point_label_xoffset`. fig, ax = plt.subplots(subplot_kw={"projection": "polar"}) ecl.plot.draw_TaylorDiagrams_base(ax=ax, std_max=2.5) ecl.plot.draw_TaylorDiagrams_metadata( taylordiagrams_metadata, ax=ax, marker_list=["o", "+", "*"], color_list=["black", "red", "green"], label_list=["1", "", "3"], legend_list=taylordiagrams_metadata["item"].to_list(), cc="cc", std="std", point_label_yoffset=[0.05, 0, 0.05], point_label_xoffset=[0.1, 0, 0], ) # %% # Modify Plotted Points # ------------------------------------ # :py:func:`easyclimate.plot.draw_TaylorDiagrams_metadata ` returns # a dictionary of plotted point and label artists. These artists can be modified after calling the plotting function. fig, ax = plt.subplots(subplot_kw={"projection": "polar"}) ecl.plot.draw_TaylorDiagrams_base(ax=ax, std_max=2.5) plot_result1 = ecl.plot.draw_TaylorDiagrams_metadata( taylordiagrams_metadata, ax=ax, cc="cc", std="std", ) items = taylordiagrams_metadata["item"].to_list() for i in range(1, 3): point = plot_result1[items[i]]["point"] point.set_marker(f"${i}$") point.set_color("r") # %% # Spacing of Axis Labels # ------------------------------------ # The parameters `x_label_pad`, `x_tickerlabel_pad`, and `y_tickerlabel_pad` can adjust # the spacing of the horizontal axis label, horizontal axis tick labels, and vertical axis tick labels. fig, ax = plt.subplots(figsize=(9, 10), subplot_kw={"projection": "polar"}) ecl.plot.draw_TaylorDiagrams_base( ax=ax, std_max=2.5, x_label_pad=0.45, x_tickerlabel_pad=20, y_tickerlabel_pad=50, ) ecl.plot.draw_TaylorDiagrams_metadata( taylordiagrams_metadata, ax=ax, cc="cc", std="std", )