Predict Future Land Cover in Google Earth Engine

Summary

In this tutorial, the presenter demonstrates how to predict land cover changes using remote sensing data and machine learning algorithms. By utilizing Sentinel satellite imagery from 2019 and 2022, along with SRTM elevation data, the speaker explains the creation of change maps and variable calculations essential for accurate predictions. The video showcases the process of analyzing land cover transitions, including urban development and agricultural shifts, while emphasizing the importance of predictive accuracy. Finally, the presenter highlights the results through visualizations and area calculations, inviting viewers to engage further with the topic.

Takeaways

• 😀 The tutorial focuses on land cover change prediction using the Random Forest algorithm for classification.
• 📊 The presenter uses Sentinel satellite data from 2019 and 2022 for their analysis.
• 🗺️ SRTM (Shuttle Radar Topography Mission) data is utilized as a predictor in the prediction process.
• 🛠️ A change map is created in QGIS to visualize and analyze land cover changes.
• 🔢 The presenter explains how to calculate and assign new values to reflect changes in land cover over the years.
• 🌳 The data from the change map is crucial for making predictions about future land use.
• 🔍 The presenter discusses the importance of setting variables to maintain consistency in predictions.
• ✅ The initial prediction model achieves an accuracy of about 90% during testing.
• 📅 The tutorial demonstrates predicting land cover changes up to 2025 based on 2022 data.
• 📈 A chart is created to visualize the area changes in different land cover classes over time.

Q & A

What is the primary focus of the video?

The video focuses on demonstrating how to perform land cover change prediction using a rendering algorithm for classification.

What years of data are used for the land cover analysis?

The analysis uses land cover data from the years 2019 and 2022.

Which data sources are mentioned in the video?

The video mentions using data from Sentinel and SRTM (Shuttle Radar Topography Mission).

What software is used to create the change map?

The change map is created using QGIS, a popular open-source geographic information system.

How does the presenter calculate changes in land cover?

The presenter calculates changes by comparing values from one year to the next and assigning new values to reflect these changes.

What is the significance of the change convergence data variable?

The change convergence data variable is crucial for making predictions about future land cover changes.

What method is used to assess the model's accuracy?

The model's accuracy is assessed by splitting the data into training and testing samples, with a reported accuracy of around 0.9.

What does the prediction for 2025 indicate about land cover?

The prediction for 2025 indicates an increase in built-up areas and a decrease in wetlands, crops, and grasslands.

How does the presenter visualize the changes in land cover over time?

The presenter visualizes changes by creating charts that display the area of different land cover types over time.

What is the overall message the presenter wants to convey regarding land cover prediction?

The presenter emphasizes the simplicity of the prediction model and encourages viewers to experiment with their own data and ask questions.