Detecting clear-cut deforestation using Landsat data : a time series analysis of remote sensing data in Covasna County, Romania between 2005 and 2015

Detta är en Master-uppsats från Lunds universitet/Institutionen för naturgeografi och ekosystemvetenskap

Sammanfattning: Forested areas represent a fundamental component of the environment. Deforestation and forest fragmentation represent a global issue mostly caused by human influence and Romania is not an exception. Nowadays forests cover approximately a third of Romania’s surface, meaning that the deforestation is a national issue. To handle this problem, a monitoring tool is necessary. The aim of the present study is to detect clear-cut deforestation using time series analysis between 2005 and 2015 in Covasna County, Romania. To achieve this objective, the analysis focuses on solving three main issues: (i) assess if clear-cut deforestation can be detected in the study area, (ii) determine the spatio-temporal distribution of the clear-cut deforestation, and (iii) confront these results with the official deforestation rates. A time series decomposition method (BFAST - Breaks For Additive Seasonal and Trend) has been used on Landsat imagery to detect forest cover changes. To identify the most suitable spectral index for detecting clear-cut deforestation, the algorithm was initially tested on a smaller test area. BFAST applied on Normalized Difference Moisture Index (NDMI) has been proved to have more consistent results than BFAST applied on Normalized Difference Vegetation Index (NDVI). This study concludes that clear-cut deforestation can be detected using BFAST algorithm considering forest type and the scale of the study area. Between 2005 and 2015, in Covasna County, Romania, 1.71% of forest cover, representing 2953 ha, was deforested. Contrary to the official deforestation rate, that mainly shows an increase of the deforestation rate for the last 10 years, the results of this study present a decrease in clear-cut deforestation between 2005 and 2015. The detection process was estimated to have an overall accuracy of 84%. Therefore, the presented method is a promising tool for monitoring clear-cut deforestation in Romania at national scale.

  HÄR KAN DU HÄMTA UPPSATSEN I FULLTEXT. (följ länken till nästa sida)