When working with complex visualizations that involve multiple layers, controlling the order of these layers is essential for creating clear and informative plots. In this post, we'll explore how to rearrange layers in Matplotlib and Seaborn plots using the zorder parameter, with an example using a weather dataset from BigQuery public datasets.

Setup

Let’s use a weather dataset from BigQuery public datasets, with the daily temperatures in New York City.

# Import libraries
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

sns.set()

%%bigquery df
# Get NYC temperatures as sample data
SELECT
    year,
    mo AS month,
    AVG(temp) AS temp
FROM `bigquery-public-data.noaa_gsod.gsod*` 
WHERE 
    wban = '94728'
    AND CAST(mo AS INT64) <= 6
GROUP BY year, mo
ORDER BY year, mo

We'll also compute the quartiles for each month:

# Compute quartiles for each month
quartiles = (
    df
    .groupby('month')
    .quantile([.25, .75])
    .reset_index()
    .pivot(index='month', columns='level_1')
    .set_axis(['low', 'high'], axis=1)
)

Reorder plot layers with z-order

When plotting multiple layers, the order in which they are drawn may not always match the sequence of the code. This can lead to unwanted visual effects, such as lines being hidden behind filled areas.

To rearrange the layers order, you can specify their z-index using the zorder parameter:

# Plot with default layers z-order
fig, ax = plt.subplots(1,2, figsize=(14,6))
sns.lineplot(data=df, x='month', y='temp', units='year', estimator=None, color='grey', alpha=.5, ax=ax[0])
ax[0].fill_between(x=quartiles.index, y1=quartiles['low'], y2=quartiles['high'], alpha=.5)
ax[0].set_title("Without z-order")

# Specify z-order
sns.lineplot(data=df, x='month', y='temp', units='year', estimator=None, color='grey', alpha=.5, ax=ax[1], zorder=1)
ax[1].fill_between(x=quartiles.index, y1=quartiles['low'], y2=quartiles['high'], alpha=.5, zorder=2)
ax[1].set_title("With z-order specified");

The zorder parameter controls the stacking order of the plot elements. A higher zorder value means the element will be drawn on top of elements with lower zorder values.

Rearrange layers with Matplotlib and Seaborn plots in Python

Setup

Reorder plot layers with z-order