An Automated Approach for Seasonal Mosaic Generation Using NDVI Time Series and Landsat 8 Data: A Case Study in a Semi-Arid Region of Brazil

This study presents an automated methodology for generating seasonal mosaics from Landsat 8 imagery using NDVI time series within the Google Earth Engine (GEE) platform. The method identifies the driest and wettest months of the year based on average NDVI values, allowing for the selection of an optimal temporal window to filter image collections and generate mosaics. The approach was applied in a semi-arid region in the western portion of Bahia, Brazil, which was the target of a geological mapping project. The region’s well-defined climatic variation favored the application of the technique. Preprocessing steps included spatial, temporal, and cloud cover filtering, which reduced the dataset to 444 images. NDVI was calculated for each image, and monthly averages were used to define the driest months (September, October, and November) and wettest months (January, February, and March). Based on this information, separate mosaics were generated for each period. The results showed clear differences between mosaics, both visually and in terms of mean NDVI values (0.16 for the dry season and 0.34 for the wet season). The dry season mosaic, characterized mainly by exposed soil, was selected for use in geological mapping. By automating steps such as optimal temporal window identification and image selection, the method significantly reduced processing time compared to manual approaches. The methodology can be easily adapted for use in other regions and applications, especially in areas with marked climatic seasonality.