I am doing Land Use Land Cover Classification in GEE using Landsat 8 Level 2, Collection 2, Tier 1 dataset. While collecting training points in the composite image, The challenge I am encountering is that urban and barren land areas appear similarly in the color composites I've tried, which were based on combinations of bands 5, 4, 3 and 7, 6, 4.
Which algorithm are you using the classification? How many samples you have? What is their spatial distribution? Which "combinations" have you tried? Where your study area is located?
Please consider adding more information to your question. As it is currently written it's nearly impossible for others to help you.
the most suitable band combination is typically the use of the Red, Green, and Blue bands (bands 4, 3, and 2 respectively).
Also, you can use some indicators to collect training samples for LULC classification. You can understand more from this paper: Land use land cover changes in the major cities of Nepal from 1990 to 2020 - ScienceDirect
When performing land use and land cover classification using satellite imagery, distinguishing between urban and barren land can indeed be challenging, especially when relying solely on visual interpretation of color composites. However, there are several techniques and additional data sources you can utilize to improve the accuracy of your classification:
1. "Additional spectral indices": Explore other spectral indices that are specifically designed to highlight urban areas or barren land. For example, the Normalized Difference Built-Up Index (NDBI) or the Normalized Difference Bareness Index (NDBaI) can help differentiate between urban and barren areas.
2. "Texture analysis": Incorporate texture analysis metrics into your classification process. Urban areas tend to have a different texture compared to barren land. Texture measures such as entropy, standard deviation, or GLCM (Gray Level Co-occurrence Matrix) features can provide valuable information for distinguishing between land cover classes.
3."Machine learning algorithms": Consider using machine learning algorithms such as Random Forest, Support Vector Machines (SVM), or Convolutional Neural Networks (CNNs) to automatically learn and differentiate between land cover classes based on a combination of spectral and contextual information.
4. "Higher-resolution imagery": If available and within your project scope, consider using higher-resolution imagery (e.g., Sentinel-2) to improve the visibility of urban features and finer details in barren land areas.
5. "Temporal analysis": Analyze temporal changes in land cover over multiple time periods to identify patterns and changes specific to urban expansion or vegetation dynamics in barren areas.
6. "Incorporate ancillary data": Utilize ancillary data sources such as population density maps, land use maps, or digital elevation models (DEM) to provide additional contextual information for classification.
7. "Data fusion": Combine multispectral imagery with other data sources such as radar data or LiDAR data to improve classification accuracy, especially in complex landscapes.
By incorporating these techniques and data sources into your land cover classification workflow, you can enhance the accuracy and reliability of distinguishing between urban and barren land areas in your study area. Experiment with different approaches and combinations to find the most suitable method for your specific project requirements.