Change Detection methods for forest expansion assessment in the last twenty years in the Mediterranean Basin

The evaluation of natural forest expansion is a crucial issue for understanding fundamental processes related to climate change and carbon uptake. The identification of forest expansion/reduction dynamics at a regional scale has been provided through multitemporal analyses of high-resolution satellite images. This study aims to compare different Change Detection (CD) techniques to assess forest expansion in the Basilicata region (Southern Italy). Landsat 5 TM image from 2003 and Landsat 8 OLI image from 2019 were used for this purpose. The CD methods implemented were NDVI Differencing and Post-Classification Comparison (PCC). Using a confusion matrix, PCC showed better performance than NDVI Differencing (Overall Accuracy of 0.85 and 0.62, respectively). The comparison of the evaluated forest expansion areas with inventory data shows the good performance of CD procedures in assessing forest land use changes. The use of satellite images and Change Detection techniques can be applied where a high temporality and a wide geographical extension are required, conditions in which traditional procedures involving field surveys are too expensive in terms of time and money. Furthermore, the use of satellite images allows for the detection of changes in the Earth’s surface in particularly inaccessible areas of the globe.