Location: Przymuszewo Forest District, Bory Tucholskie, Poland
Software: ERDAS Imagine
Data sources: Sentinel-2 satellite imagery (2015, 2019)
This project involved a detailed analysis of land cover changes in the Bory Tucholskie region, which suffered extensive forest damage due to a severe windstorm in 2017. The main objective was to detect deforestation and forest succession between 2015 and 2019 using satellite imagery and remote sensing techniques.
Key tasks:
- Data preprocessing: Created multispectral composites from Sentinel-2 imagery (2015 and 2019) and clipped the images to the forest district boundary.
- Land cover classification:
Performed supervised classification using training samples based on four land cover classes: forest, agricultural/low vegetation areas, built-up areas, and water bodies. The classification was conducted using the maximum likelihood algorithm, and results were aggregated into two main categories: forest and non-forest.
- Change detection – qualitative approach:
Applied post-classification comparison to identify areas of deforestation, afforestation, and unchanged land. Compared results with Hansen et al.’s Global Forest Change dataset.
- Change detection – quantitative approach:
Calculated NDVI values for both years and generated an NDVI difference map. Applied thresholding via a graphical model in ERDAS Model Maker to classify areas with significant vegetation loss or gain.
- Comparison of methods:
Compared both change detection approaches to highlight consistent results as well as discrepancies caused by different vegetation dynamics (e.g., coniferous forest replaced by high-NDVI grassland).
Outcome:
The project successfully identified environmental changes caused by natural disaster and vegetation dynamics, showcasing the use of remote sensing and GIS techniques in environmental monitoring and land management.