如何优化近0值占比高的等高线图并实现类瀑布图效果？

To fix your contour plot and achieve the desired waterfall-style visualization with a blue background and prominent meaningful data, apply these adjustments:

Key Fixes: #

• Use SymLogNorm instead of Normalize or LogNorm to handle zero values while preserving visibility of 1e-7 range data.
• Replace zero values with a tiny positive number to avoid log scaling errors.
• Define custom color levels to focus on your critical value ranges.
• Combine contourf (filled background) and contour (top lines) for the waterfall effect.

Modified Code: #

import numpy as np
import matplotlib.pyplot as plt
from matplotlib.colors import SymLogNorm

X, Y = np.meshgrid(frequencies, speeds)
Z = freq_speed_grid.to_numpy()

# Replace zeros with a small positive value to enable log scaling
Z[Z == 0] = 1e-8

plt.figure(figsize=(10, 5))

# SymLogNorm handles zero/near-zero values and log scales for larger data
norm = SymLogNorm(linthresh=1e-7, vmin=1e-8, vmax=Z.max(), base=10)

# Custom levels to emphasize the 1e-7 range and meaningful data
levels = np.concatenate([
    np.linspace(1e-7, 1e-6, 10),  # Dense levels for low-value range
    np.linspace(1e-6, Z.max(), 40)  # Sparser levels for higher values
])

# Filled contour for blue background (low values)
plt.contourf(X, Y, Z, cmap="jet", norm=norm, levels=levels, extend="min")

# Overlay white contour lines to highlight data boundaries
plt.contour(X, Y, Z, colors="white", linewidths=0.5, levels=levels)

plt.colorbar(label="Magnitude")
plt.xlabel("Frequency (Hz)")
plt.ylabel("Speed (RPM)")
plt.show()

Explanation: #

• SymLogNorm: Maps zero/near-zero values to the blue background via linear scaling near linthresh=1e-7, while using log scaling for larger values to preserve detail in meaningful data.
• Zero Replacement: Prevents LogNorm's all-white issue by giving zeros a non-zero value that maps to the lowest color in the colormap.
• Custom Levels: Concentrates more levels around 1e-7 to enhance color differentiation in that critical range, making subtle variations visible.
• Contour Overlay: White lines on top of the filled contour mimic the waterfall plot's distinct, easy-to-read data boundaries.

内容的提问来源于stack exchange，提问作者Firas Bejar