Integration of Spatial and Fractal Analysis for Evaluating Urban Green Areas

Abstract

Accurate assessment of urban green areas is essential for enhancing livability and guiding sustainable urban planning. This study investigates green space distribution in Samsun, Turkey, by integrating spatial and fractal analyses. Green areas in 77 neighborhoods across the Atakum, Ilkadim, and Canik districts were identified using the normalized difference vegetation index (NDVI) derived from Sentinel-2A/2B imagery (July-October 2023). To extract green areas, both Otsu's automatic and manual thresholding methods were applied. Manual thresholding demonstrated higher classification accuracy in heterogeneous urban contexts, based on spectral discrimination index and confusion matrix evaluation. Spatial assessment employed three indicators: Per Capita Green Space (PCGS), Urban Green Space Index (UGSI), and Urban Green Density Index (UGDI). Fractal dimension and lacunarity index were calculated using the box-counting and gliding box methods, respectively, to assess morphological structure. To synthesize these indicators, both equal-weighted overlay and principal component analysis (PCA) were applied. PCA-based aggregation (65.8% of total variance explained by the first component) was adopted to compute the urban green area service level (UGASL). UGASL scores were classified into five levels: very high, high, medium, low, and very low. In Atakum, 36.8% of neighborhoods were medium, 31.6% low, and 31.6% very low. In Ilkadim, 6.7% were high, 20% medium, 13.3% low, and 60% very low. In Canik, 7.7% were very high, 7.7% high, 30.8% medium, 15.4% low, and 38.4% very low. The study highlights intra-urban disparities and demonstrates that combining spatial and fractal metrics via PCA provides a robust, scalable framework for equitable and sustainable green infrastructure planning.