Volume 28, Issue 4-1, December 2024
PHENOLOGY ANALYSIS FOR DETECTION OF VEGETATION CHANGES BASED ON LANDSAT 8 IMAGES IN NATURE PARK KOPAČKI RIT, CROATIA
Authors: Dorijan Radočaj, Ivan Plaščak, Mladen Jurišić, Ivana Majić, Siniša Ozimec, Ankica Sarajlić, Vlatko Rožac
Abstract: This study proposed a method for detecting vegetation changes and establishing geospatial management zones based on the 10-year phenology analysis using normalized difference vegetation index (NDVI) long-term trends from Landsat 8 multispectral imagery in Nature Park Kopački rit. The main components of the proposed method include phenology analysis and NDVI anomaly detection supported by unsupervised k-means classification of vegetation management zones. The reference monthly NDVI values (2013–2019) with three test years (2020–2022) strongly indicated very high heterogeneity in vegetation activity. A 100 m spatial resolution and a monthly temporal resolution were used. The results of unsupervised k-means classification in five vegetation activity classes indicated that three of these classes have considerably high negative NDVI anomalies, covering 64.1% of the study area. While the proposed method ensures the detection of vegetation changes and vegetation activity zones, a comprehensive field observation is required to determine the potential environmental and/or anthropogenic causes. However, the proposed approach significantly reduces the need for extensive fieldwork, allowing biologists to focus their efforts on areas with detected abnormal vegetation activity.
Keywords: vegetation activity, “npphen” package, management zones, phenology analysis, normalized difference vegetation index (NDVI), k-means classification
Article info:
Received: May 2, 2024| Revised: September 27, 2024| Accepted: October 22, 2024