End-to-End Earthquake Data Pipeline with Azure and Databricks by Edidiong EsuEnd-to-End Earthquake Data Pipeline with Azure and Databricks by Edidiong Esu

End-to-End Earthquake Data Pipeline with Azure and Databricks

Edidiong Esu

Completed work

Data Analyst

Data Engineer

Microsoft Power BI

Python

Data

End-to-End Earthquake Data Engineering & Analytics Pipeline with Azure, Databricks & Power BI

Navigation / Quick Access

Quickly move to the section you are interested in by clicking on the appropriate link:

Reproducing Project (long section)

Dashboard

Overview

Every day, hundreds of earthquakes shake the earth — some minor, others devastating. Understanding where, when, and how they occur is crucial for monitoring natural hazards, informing infrastructure planning, and protecting communities.

This project extracts and analyzes real-time earthquake data from the United States Geological Survey (USGS), an agency that tracks seismic activity around the globe. The data is ingested daily via Azure Data Factory, transformed in Azure Databricks using the medallion architecture (bronze → silver → gold), stored in Microsoft Fabric Lakehouse, and visualized using Power BI.

The pipeline demonstrates how raw JSON data from a public API can be converted into structured, trusted insights using modern, cloud-native tools. By the end of the pipeline, users can explore:

🌍 Earthquake hotspots by country and region

📈 Magnitude trends over time

🚨 Significant seismic events by signal strength

🕒 Time-based patterns of earthquake activity

This project showcases how scalable data engineering workflows can power decision-ready dashboards, turning global sensor data into actionable intelligence for analysts, researchers, and the public.

Project Objective

✅ Automate daily ingestion of global earthquake data from the USGS API using Azure Data Factory

✅ Transform and enrich raw data using Azure Databricks with a medallion architecture (bronze → silver → gold)

✅ Store trusted and structured data in Microsoft Fabric Lakehouse

✅ Build interactive Power BI dashboards that uncover patterns, trends, and anomalies in global seismic activity

Project Architecture

This architecture illustrates the end-to-end data pipeline used in this project, leveraging Azure and Microsoft Fabric services to move from raw ingestion to visual insights.

🔄 End-to-End Pipeline Flow

🔁 Ingestion – Azure Data Factory

Daily earthquake data is ingested from the USGS API.

Azure Data Factory orchestrates the process and stores the data in Azure Data Lake.

⚙️ Transformation – Azure Databricks (Medallion Architecture)

Data is processed through three structured layers:

Bronze Layer: Raw ingestion and flattening

Silver Layer: Cleansing, filtering, standardization

Gold Layer: Aggregated and enriched for reporting

🏠 Storage – Microsoft Fabric Lakehouse

The gold-layer data is loaded into Microsoft Fabric Lakehouse for scalable storage and advanced analytics.

📊 Visualization – Power BI

Fabric Lakehouse feeds directly into Power BI, enabling dynamic dashboards and reports for stakeholders.

✅ This architecture ensures a reliable, scalable, and analytics-ready pipeline from API to dashboard.

Dataset

🌍 Source: USGS Earthquake API

This project collects seismic data from the United States Geological Survey (USGS) Earthquake API, which provides detailed information about global earthquake events.

API Endpoint: https://earthquake.usgs.gov/fdsnws/event/1/

Data Format: GeoJSON

Ingestion: Daily via Azure Data Factory

Dynamic Parameters:

starttime: set dynamically during ingestion

endtime: optional, defaults to the same as starttime

📘 API Documentation

Gold Layer Schema (Final Output)

The final dataset is produced in the gold layer after cleaning, enrichment, and transformation in Databricks. This output is ready for analytics or visualization.

|-- id: string (nullable = true)
|-- longitude: double (nullable = true)
|-- latitude: double (nullable = true)
|-- elevation: double (nullable = true)
|-- title: string (nullable = true)
|-- place_description: string (nullable = true)
|-- sig: long (nullable = true)
|-- mag: double (nullable = true)
|-- magType: string (nullable = true)
|-- time: timestamp (nullable = true)
|-- updated: timestamp (nullable = true)
|-- country_code: string (nullable = true)
|-- sig_class: string (nullable = false)

Reproducing Project

This project uses Azure Data Factory to move data through Azure Databricks notebooks (bronze, silver, gold layers). Follow the steps below to set up the environment.

1. Create a Free Azure Account

If you don’t already have one, start by creating a free Azure account:

🔗 https://azure.microsoft.com/free

You’ll get free credits and access to services like Data Factory and Databricks.

2. Create a Resource Group

A resource group organizes related Azure resources.

In Azure Portal, search for "Resource groups"

Click Create

Choose a region and give it a name (e.g., data-project-rg)

3. Set Up Azure Data Factory (ADF)

In Azure Portal, search “Data Factory”

Click Create

Select your resource group, choose a name (e.g., datafactory-dev), and location

Leave version as V2

Once deployed, click Author & Monitor to open the ADF Studio

📘 Create Data Factory

4. Create a Databricks Workspace

In Azure Portal, search “Databricks”

Click Create

Fill in:

Workspace name (e.g., databricks-dev)

Same region as ADF

Pricing tier: Standard. There is a trial tier that is limited for 14 days also

After deployment, click Launch Workspace to open the Databricks UI

📘 Create Azure Databricks Workspace

5. Create a Databricks Cluster & Notebook

Inside the Databricks workspace:

Go to Compute → Click Create Cluster

Runtime: 11.3 LTS for mInimal Configuration. Choose cluster to turn off after 30 mins of inactivity to save cost.

Keep other settings default for development

Go to Workspace → Your username → Click New > Notebook

Name: bronze_ingest

Language: Python or PySpark

6. Generate Access Token (for ADF to Databricks)

In Databricks, click your profile (top-right) → User Settings

Go to Access Tokens → Generate New Token

Copy the token (save it securely)

📘 Access Tokens in Databricks

7. Connect ADF to Databricks

Back in ADF Studio:

Go to Manage > Linked Services > + New

Choose Azure Databricks

Fill in:

Workspace URL (from Azure portal)

Access token

Cluster ID (found in Databricks > Compute > cluster > URL)

Now ADF can run your Databricks notebooks as activities in a pipeline.

📘 Run Databricks from ADF

8. 🧪 Configure Earthquake Pipeline in ADF Using Databricks Notebooks

After setting up the Azure environment and connecting to Databricks, the next step is to configure the earthquake pipeline using a series of Databricks notebooks that follow the medallion architecture: Bronze → Silver → Gold.

Each notebook runs as an activity in the ADF pipeline, and parameters are passed between them dynamically.

📁 ADF Pipeline Overview

The ADF pipeline orchestrates the daily execution of the Databricks notebooks. Create a pipeline in adf called pl-earthquake. Below is the structure of the pipeline:

🟫 Bronze Notebook

The Bronze Notebook ingests raw earthquake data and stores it in json format in the Bronze layer of the lakehouse.

🔧 Base Parameters

The notebook takes two date parameters — start_date and end_date — which are dynamically generated in ADF using the following expressions:

{
  "start_date": "@formatDateTime(addDays(utcNow(), -1), 'yyyy-MM-dd')",
  "end_date": "@formatDateTime(utcNow(), 'yyyy-MM-dd')"
}

At the end of this step, the raw earthquake data is stored in JSON format in the Bronze layer of Azure Data Lake Gen2, ready for further processing in Databricks.

⚙️ Silver Notebook

The Silver Notebook notebook processes the JSON data from the Bronze layer by flattening nested fields (like coordinates and properties), renaming columns, and handling missing values.

🔧 Base Parameters

{
  "bronze_params": "@string(activity('Bronze Notebook').output.runOutput)"
}

The cleaned and flattened data is saved in parquet format in the Silver layer of Azure Data Lake Gen2, making it ready for enrichment.

🏅 Gold Notebook

The Gold Notebook enriches the Silver data by assigning a country_code to each earthquake event using reverse geocoding, and classifies each event's significance (sig) as Low, Moderate, or High to support downstream analytics.

🔧 Base Parameters

{
  "bronze_params": "@string(activity('Bronze Notebook').output.runOutput)",
  "silver_params": "@string(activity('Silver Notebook').output.runOutput)"
}

The enriched dataset, now with country codes and significance classifications, is stored in Delta format in the Gold layer and made available to Microsoft Fabric Lakehouse for downstream analytics,

Triggers

A daily Trigger is created and time set to run the pipeline daily. This is after debug run has been done and everything confirmed to work fine

9. 🏞️ Moving Data to Microsoft Fabric Lakehouse

After enrichment in the Gold notebook including the addition of fields like `country_code, The output files are passed into a Fabrics Notebook for final processing

The Fabrics Notebook performs the following:

Merges partitioned output files

Applies any final transformations (e.g., standardizing country_name)

Outputs a clean, unified Delta table in the Microsoft Fabric Lakehouse

This structured table serves as the final dataset for reporting and analytics.

Key benefits:

Stored in Delta format within the OneLake architecture

Optimized for low-latency querying

Natively available to Power BI through Direct Lake mode

Dashboard

Visualizing in Power BI

The data in the Fabric Lakehouse is visualized using Power BI to deliver insights on seismic activity trends, frequency, significance distribution, and geographic patterns.

Dashboards include:

Top Countries with Earthquakes

Magnitude distribution charts and the Earthquake with the most impact

Geographic mapping of earthquake events

Severity class breakdown (Low / Moderate / High) and lots more!

Looking Closer:

The pipeline has been paused to save costs, so the Visualization link is currently unavailable for public consumption.

Contact

Please reach out to me on LinkedIn for thoughts and/or issues encountered during the reproduction of this project. Let's chat! ⭐.

Happy Coding! 💻

Like this project

Completed work

Posted Sep 15, 2025

Built an end-to-end earthquake data pipeline using Azure, Databricks, and Power BI.

Likes

Views

End-to-End Earthquake Data Pipeline with Azure and Databricks

Edidiong Esu

Completed work

Data Analyst

Data Engineer

Microsoft Power BI

Python

Data

End-to-End Earthquake Data Engineering & Analytics Pipeline with Azure, Databricks & Power BI

Navigation / Quick Access

Quickly move to the section you are interested in by clicking on the appropriate link:

Reproducing Project (long section)

Dashboard

Overview

The pipeline demonstrates how raw JSON data from a public API can be converted into structured, trusted insights using modern, cloud-native tools. By the end of the pipeline, users can explore:

🌍 Earthquake hotspots by country and region

📈 Magnitude trends over time

🚨 Significant seismic events by signal strength

🕒 Time-based patterns of earthquake activity

This project showcases how scalable data engineering workflows can power decision-ready dashboards, turning global sensor data into actionable intelligence for analysts, researchers, and the public.

Project Objective

✅ Automate daily ingestion of global earthquake data from the USGS API using Azure Data Factory

✅ Transform and enrich raw data using Azure Databricks with a medallion architecture (bronze → silver → gold)

✅ Store trusted and structured data in Microsoft Fabric Lakehouse

✅ Build interactive Power BI dashboards that uncover patterns, trends, and anomalies in global seismic activity

Project Architecture

This architecture illustrates the end-to-end data pipeline used in this project, leveraging Azure and Microsoft Fabric services to move from raw ingestion to visual insights.

🔄 End-to-End Pipeline Flow

🔁 Ingestion – Azure Data Factory

Daily earthquake data is ingested from the USGS API.

Azure Data Factory orchestrates the process and stores the data in Azure Data Lake.

⚙️ Transformation – Azure Databricks (Medallion Architecture)

Data is processed through three structured layers:

Bronze Layer: Raw ingestion and flattening

Silver Layer: Cleansing, filtering, standardization

Gold Layer: Aggregated and enriched for reporting

🏠 Storage – Microsoft Fabric Lakehouse

The gold-layer data is loaded into Microsoft Fabric Lakehouse for scalable storage and advanced analytics.

📊 Visualization – Power BI

Fabric Lakehouse feeds directly into Power BI, enabling dynamic dashboards and reports for stakeholders.

✅ This architecture ensures a reliable, scalable, and analytics-ready pipeline from API to dashboard.

Dataset

🌍 Source: USGS Earthquake API

This project collects seismic data from the United States Geological Survey (USGS) Earthquake API, which provides detailed information about global earthquake events.

API Endpoint: https://earthquake.usgs.gov/fdsnws/event/1/

Data Format: GeoJSON

Ingestion: Daily via Azure Data Factory

Dynamic Parameters:

starttime: set dynamically during ingestion

endtime: optional, defaults to the same as starttime

📘 API Documentation

Gold Layer Schema (Final Output)

The final dataset is produced in the gold layer after cleaning, enrichment, and transformation in Databricks. This output is ready for analytics or visualization.

|-- id: string (nullable = true)
|-- longitude: double (nullable = true)
|-- latitude: double (nullable = true)
|-- elevation: double (nullable = true)
|-- title: string (nullable = true)
|-- place_description: string (nullable = true)
|-- sig: long (nullable = true)
|-- mag: double (nullable = true)
|-- magType: string (nullable = true)
|-- time: timestamp (nullable = true)
|-- updated: timestamp (nullable = true)
|-- country_code: string (nullable = true)
|-- sig_class: string (nullable = false)

Reproducing Project

This project uses Azure Data Factory to move data through Azure Databricks notebooks (bronze, silver, gold layers). Follow the steps below to set up the environment.

1. Create a Free Azure Account

If you don’t already have one, start by creating a free Azure account:

🔗 https://azure.microsoft.com/free

You’ll get free credits and access to services like Data Factory and Databricks.

2. Create a Resource Group

A resource group organizes related Azure resources.

In Azure Portal, search for "Resource groups"

Click Create

Choose a region and give it a name (e.g., data-project-rg)

3. Set Up Azure Data Factory (ADF)

In Azure Portal, search “Data Factory”

Click Create

Select your resource group, choose a name (e.g., datafactory-dev), and location

Leave version as V2

Once deployed, click Author & Monitor to open the ADF Studio

📘 Create Data Factory

4. Create a Databricks Workspace

In Azure Portal, search “Databricks”

Click Create

Fill in:

Workspace name (e.g., databricks-dev)

Same region as ADF

Pricing tier: Standard. There is a trial tier that is limited for 14 days also

After deployment, click Launch Workspace to open the Databricks UI

📘 Create Azure Databricks Workspace

5. Create a Databricks Cluster & Notebook

Inside the Databricks workspace:

Go to Compute → Click Create Cluster

Runtime: 11.3 LTS for mInimal Configuration. Choose cluster to turn off after 30 mins of inactivity to save cost.

Keep other settings default for development

Go to Workspace → Your username → Click New > Notebook

Name: bronze_ingest

Language: Python or PySpark

6. Generate Access Token (for ADF to Databricks)

In Databricks, click your profile (top-right) → User Settings

Go to Access Tokens → Generate New Token

Copy the token (save it securely)

📘 Access Tokens in Databricks

7. Connect ADF to Databricks

Back in ADF Studio:

Go to Manage > Linked Services > + New

Choose Azure Databricks

Fill in:

Workspace URL (from Azure portal)

Access token

Cluster ID (found in Databricks > Compute > cluster > URL)

Now ADF can run your Databricks notebooks as activities in a pipeline.

📘 Run Databricks from ADF

8. 🧪 Configure Earthquake Pipeline in ADF Using Databricks Notebooks

Each notebook runs as an activity in the ADF pipeline, and parameters are passed between them dynamically.

📁 ADF Pipeline Overview

The ADF pipeline orchestrates the daily execution of the Databricks notebooks. Create a pipeline in adf called pl-earthquake. Below is the structure of the pipeline:

🟫 Bronze Notebook

The Bronze Notebook ingests raw earthquake data and stores it in json format in the Bronze layer of the lakehouse.

🔧 Base Parameters

The notebook takes two date parameters — start_date and end_date — which are dynamically generated in ADF using the following expressions:

{
  "start_date": "@formatDateTime(addDays(utcNow(), -1), 'yyyy-MM-dd')",
  "end_date": "@formatDateTime(utcNow(), 'yyyy-MM-dd')"
}

At the end of this step, the raw earthquake data is stored in JSON format in the Bronze layer of Azure Data Lake Gen2, ready for further processing in Databricks.

⚙️ Silver Notebook

The Silver Notebook notebook processes the JSON data from the Bronze layer by flattening nested fields (like coordinates and properties), renaming columns, and handling missing values.

🔧 Base Parameters

{
  "bronze_params": "@string(activity('Bronze Notebook').output.runOutput)"
}

The cleaned and flattened data is saved in parquet format in the Silver layer of Azure Data Lake Gen2, making it ready for enrichment.

🏅 Gold Notebook

🔧 Base Parameters

{
  "bronze_params": "@string(activity('Bronze Notebook').output.runOutput)",
  "silver_params": "@string(activity('Silver Notebook').output.runOutput)"
}

The enriched dataset, now with country codes and significance classifications, is stored in Delta format in the Gold layer and made available to Microsoft Fabric Lakehouse for downstream analytics,

Triggers

A daily Trigger is created and time set to run the pipeline daily. This is after debug run has been done and everything confirmed to work fine

9. 🏞️ Moving Data to Microsoft Fabric Lakehouse

After enrichment in the Gold notebook including the addition of fields like `country_code, The output files are passed into a Fabrics Notebook for final processing

The Fabrics Notebook performs the following:

Merges partitioned output files

Applies any final transformations (e.g., standardizing country_name)

Outputs a clean, unified Delta table in the Microsoft Fabric Lakehouse

This structured table serves as the final dataset for reporting and analytics.

Key benefits:

Stored in Delta format within the OneLake architecture

Optimized for low-latency querying

Natively available to Power BI through Direct Lake mode

Dashboard

Visualizing in Power BI

The data in the Fabric Lakehouse is visualized using Power BI to deliver insights on seismic activity trends, frequency, significance distribution, and geographic patterns.

Dashboards include:

Top Countries with Earthquakes

Magnitude distribution charts and the Earthquake with the most impact

Geographic mapping of earthquake events

Severity class breakdown (Low / Moderate / High) and lots more!

Looking Closer:

The pipeline has been paused to save costs, so the Visualization link is currently unavailable for public consumption.

Contact

Please reach out to me on LinkedIn for thoughts and/or issues encountered during the reproduction of this project. Let's chat! ⭐.

Happy Coding! 💻

Like this project

Completed work

Posted Sep 15, 2025

Built an end-to-end earthquake data pipeline using Azure, Databricks, and Power BI.

Likes

Views