LAUSR

Urban forest and vegetation conditions are an important variable in urban ecosystem management decision-making. However, it is difficult to evaluate and monitor solely on the basis of field measurements. Remote sensing technologies can greatly contribute to the faster extraction and mapping of vegetation health status indicators, on the basis of which agronomy and forestry experts can draw conclusions about the condition of urban vegetation in larger areas. A new remote sensing-based urban forest and vegetation cover monitoring framework is presented and applied to a case study of the city of Zagreb, Croatia. In this study, Sentinel-2 multi-temporal imagery was used to derive and analyze the current state of urban forest cover. Vegetation indices (NDVI, RVI, and GRVI) were calculated. K-means unsupervised classification of the vegetation indices was conducted. In this way, the dimensionality of the vegetation indices was reduced, while all the data contained in it were used to represent their graded values. Vegetation that was in a poor condition stood out better that way. Finally, PCA-based change detection was performed on the vegetation indices graded values, and a map of change was produced. These results need to be interpreted and validated by foresters and agronomists in further research.