Spatial Join of Two Dataframes in PySpark

What will you learn?

In this tutorial, you will learn how to execute a spatial join on two dataframes using PySpark. By combining the attributes of these dataframes based on the spatial relationship between their geometries, you can enrich your data analysis and gain valuable insights.

Introduction to the Problem and Solution

Imagine having two separate datasets with geographical information that you want to merge based on their spatial proximity. This is where a spatial join comes into play. By leveraging PySpark’s capabilities for handling spatial operations efficiently, we can seamlessly combine these datasets and extract meaningful correlations.

To tackle this task effectively, we will: – Import essential libraries including PySpark and GeoPandas. – Create a Spark session for our application. – Load the datasets into PySpark dataframes. – Convert these dataframes into GeoPandas GeoDataFrames to perform the spatial join operation. – Utilize GeoPandas’ sjoin() function to carry out the spatial join based on geometric intersections. – Optionally convert the resulting joined GeoDataFrame back to a PySpark DataFrame if needed.


# Import necessary libraries
from pyspark.sql import SparkSession
from pyspark.sql.functions import broadcast
import geopandas as gpd

# Create a Spark session
spark = SparkSession.builder \
    .appName("SpatialJoinExample") \

# Load the dataframes into PySpark dataframes (df1 and df2)
df1 ="csv").option("header", "true").load("path_to_dataframe1.csv")
df2 ="csv").option("header", "true").load("path_to_dataframe2.csv")

# Convert the PySpark dataframes to GeoPandas GeoDataFrames (gdf1 and gdf2)
gdf1 = gpd.GeoDataFrame(df1.toPandas(), geometry=gpd.points_from_xy(df1.longitude, df1.latitude))
gdf2 = gpd.GeoDataFrame(df2.toPandas(), geometry=gpd.points_from_xy(df2.longitude, df2.latitude))

# Perform Spatial Join using GeoPandas
result_gdf = gpd.sjoin(gdf1, gdf2, how="inner", op="intersects")

# Convert the resulting GeoDataFrame back to a PySpark DataFrame if needed
result_df = spark.createDataFrame(result_gdf)

# Show or save the final result as required

# Copyright PHD

Remember to replace path_to_dataframe1.csv and path_to_dataframe2.csv with your actual file paths.


In this code snippet: – We start by importing necessary libraries such as pyspark and geopandas. – A Spark session is created for our application. – The CSV files are loaded into two separate PySpark dataframes. – These dataframes are converted into Geopandas GeoDataFrames for conducting the spatial join operation. – The GeoPandas library’s sjoin() function is used for executing the spatial join based on intersecting point geometries. – If required, the resulting joined Geodataframe is transformed back into a Pyspark DataFrame.

    How does a spatial join differ from other types of joins?

    A spatial join operates based on geometric relationships like intersecting shapes rather than traditional column or key matches in other joins.

    Can I perform a spatial join with non-spatial data?

    No, meaningful spatial joins require geospatial information or geometry columns in both datasets.

    What are some common geometric operations used in spatial joins?

    Common operations include intersects, contains, within when performing spatial joins.

    Is it possible to optimize performance when working with large datasets?

    Yes, optimizing performance involves proper partitioning of data and utilizing broadcast variables where applicable.

    How do I handle duplicate column names during merging?

    You can manage conflicting column names by specifying suffixes using appropriate parameters in merge functions.


    Mastering spatial joins in PySpark opens up avenues for insightful analysis of location-based datasets by leveraging geographic relationships between entities. With tools like PySpark and GeoPandas, handling such tasks efficiently at scale becomes achievable.

    Leave a Comment