A 25-year Landsat-derived environmental observation dataset for Narayanganj, Bangladesh, 2001–2025

This repository contains a 25-year point-based environmental observation dataset for Narayanganj District, Bangladesh, derived from Landsat Collection 2 Tier 1 Level-2 imagery processed in Google Earth Engine. The dataset comprises 3,605,983 valid pixel observations extracted from 1,415 Landsat 5, 7, 8, and 9 scenes acquired between 2001 and 2025 at 5,000 fixed sampling locations. The dataset is distributed as annual CSV files containing calibrated surface reflectance bands (Blue, Green, Red, Near Infrared, SWIR1, and SWIR2), thermal measurements, land surface temperature (°C), water occurrence percentage, QA_PIXEL values, cloud cover, acquisition metadata, solar geometry, WRS path/row, geographic coordinates, and Landsat mission information. All observations were generated using a reproducible Google Earth Engine workflow that applies mission-specific radiometric scaling, quality masking, band harmonization, and fixed-point sampling. In addition to the annual datasets, the repository includes the complete Google Earth Engine extraction script, enabling transparent reproduction of the data generation process and adaptation to other regions or time periods. The dataset supports long-term environmental monitoring, urban heat island studies, land-cover and land-surface temperature analysis, machine learning, and other remote sensing applications.