Analyzing Land Surface Temperature(LST) with Google Earth Engine: Insights into Urban Heat Island Effects in Bangalore

Introduction:

Land Surface Temperature (LST) is a critical variable for studying the Earth's surface energy balance and understanding climate patterns. Leveraging Google Earth Engine (GEE), we can efficiently analyze LST over time using satellite-derived data. This blog demonstrates how to perform an LST time series analysis with Landsat data.

This script performs Land Surface Temperature (LST) analysis over a specified area of interest (AOI) using Landsat 8 Surface Reflectance products. Below is a detailed breakdown of its components:

Methodologies

Load Landsat 8 data:

To begin the analysis, Landsat 8 Surface Reflectance data is loaded into Google Earth Engine. This dataset provides detailed multispectral imagery essential for LST calculations.

Steps:

• Define the Area of Interest (AOI), typically a specific region such as Bangalore Urban.

• Load the Landsat 8 data collection for the desired time range.

• Apply filters to select only relevant imagery, such as removing scenes with excessive cloud cover or those outside the AOI.

Mask Invalid pixels:

Data quality is vital for accurate LST analysis. Cloud cover, shadows, and missing pixels can introduce inaccuracies. In this step, invalid pixels are masked out using GEE’s built-in functions.

Steps:

• Apply the cloud mask using the QA band in Landsat 8, which flags pixels affected by clouds, shadows, or aerosols.

• Exclude these flagged pixels from the analysis, ensuring the dataset consists only of valid observations.

Calculate LST in Celsius:

The Landsat 8 data includes thermal bands that provide at-sensor brightness temperature. This temperature must be converted into LST in Celsius for meaningful interpretation.

Steps:

• Use the thermal bands (Band 10 or Band 11) to retrieve brightness temperature.

• Convert the temperature from Kelvin to Celsius using the formula: LST(∘C)=Brightness Temperature (K)−273.15LST(∘C)=Brightness Temperature (K)−273.15

• Apply surface emissivity correction to account for differences in land cover types.

Calculate yearly Mean LST:

Yearly averages help identify seasonal and long-term trends in LST. Aggregating data over time reduces noise and highlights meaningful patterns.

Steps:

• Group the LST data by year.

• Calculate the mean LST for each year by averaging the temperature values across all valid pixels within the AOI.

• Store these yearly mean values for visualization and further analysis.

Visualize yearly mean LST:

Visualization is crucial for interpreting trends and patterns in LST data. Google Earth Engine provides tools for creating clear, insightful visual outputs.

Steps:

• Generate a color-coded map overlay showing the spatial distribution of mean LST values for each year.

• Create line charts or bar graphs to represent yearly LST trends, helping to identify anomalies, spikes, or steady patterns.

• Export visualizations in a publication-ready format for reports or presentations.

Export yearly mean LST to your local machine:

This step involves exporting the yearly mean Land Surface Temperature (LST) values to your local machine for further analysis or presentation purposes. The provided script automates the export process using Google Earth Engine's (GEE) built-in Export.image.toDrive() function. Here's an explanation of the key components and steps:

1. Iterating Over Yearly LST Data

The script loops through a list of yearly mean LST images (yearlyMeanLST). Each iteration retrieves the LST data for a specific year, allowing the export process to handle each year's dataset individually.

• The getInfo() method fetches the metadata, such as the year, for each image in the list.

• Each image is accessed using yearlyImagesList.get(i), where i represents the index of the current image.

2. Preparing the Export Settings

The export settings are defined within the Export.image.toDrive() function:

• Image Selection: The script selects the specific band ('LST_C') from the image that represents LST values in Celsius.

• Description: The exported file is named dynamically based on the year, following the format 'Mean_LST_Raw_' + year.

• Folder: All exported files are stored in a folder named 'EarthEngineExports' on Google Drive.

• Scale: The spatial resolution of the output is set to 30 meters, matching the resolution of Landsat 8 data.

• Region: The export is limited to the defined Area of Interest (AOI), ensuring only relevant data is saved.

• Maximum Pixels: The maxPixels parameter allows for handling large datasets by setting an upper limit (1e13 pixels).

• File Format: The exported files are saved in GeoTIFF format, which is compatible with most GIS tools.

3. Automating Exports for All Years

The loop processes and exports each year's data one by one, making it efficient for long time series analysis. Each file is saved separately, providing a clear, year-wise dataset that can be easily referenced.

4. Accessing the Exported Data

Once the script is executed:

• The GeoTIFF files for each year are saved in your Google Drive under the specified folder (EarthEngineExports).

• These files can be downloaded to your local machine for visualization or further analysis using tools such as QGIS, ArcGIS, or Python.

The Output/Result:

From the chart it is clear that over the years, the LST values generally stay within the range of approximately 28°C to 44°C. There doesn’t appear to be a consistent increasing or decreasing trend, suggesting no significant long-term warming or cooling trend over the study period. Some years show sharp spikes or troughs that deviate from the general seasonal pattern, which could indicate: Extreme weather events (e.g., heatwaves or cold spells) or it might be data noise or incomplete cloud masking. Being a densely urbanized area, Bangalore Urban likely experiences an Urban Heat Island (UHI) effect, where urban surfaces (concrete, asphalt) absorb and retain heat, causing elevated LST values compared to surrounding rural areas. The consistently high peak temperatures may reflect this urban heating phenomenon.

Conclusion:

The analysis of Land Surface Temperature using Google Earth Engine offers valuable insights into urban climate dynamics and environmental challenges. Despite the absence of a long-term warming trend, the consistently high LST values in Bangalore Urban highlight the profound impact of urbanization and the Urban Heat Island effect. These findings underscore the need for sustainable planning and adaptive measures to mitigate urban heat impacts and promote environmental resilience. GEE proves to be a powerful tool for large-scale geospatial analysis, facilitating actionable insights for researchers and policymakers.