TY  - EJOU
AU  - Eren, Esra 
AU  - Özelkan, Emre 

TI  - Multi-Scale Spatial Analysis of Urban Heat Island Dynamics: Linking Land Surface Temperature and Urban Density in Çanakkale (Türkiye)
T2  - Revue Internationale de Géomatique

PY  - 2026
VL  - 35
IS  - 1
SN  - 2116-7060

AB  - The Urban Heat Island (UHI) effect is a significant consequence of urbanization that shapes Land Surface Temperature (LST) patterns and thermal variability through the complex interactions of urban morphology, surface characteristics, and built environment configuration. This study investigates the multi-scale relationship between urban density indicators and LST in the coastal city of Çanakkale, Türkiye, over a 14-year period (2010–2023). Utilizing Landsat thermal data, the research evaluates six indicators, including Building Coverage Ratio (BCR), Floor Area Ratio (FAR), population density, and open space density, across four spatial scales: 30 m grid, building block, neighborhood, and district. Results identify the Building Block scale as the optimal resolution. While the Neighborhood scale initially exhibited higher explanatory power (<math id="mml-ieqn-1"><msubsup><mi>R</mi><mrow><mi>a</mi><mi>d</mi><mi>j</mi></mrow><mrow><mn>2</mn></mrow></msubsup></math> = 0.447), Leave-One-Out Cross-Validation (LOOCV) revealed this as an artifact of spatial overfitting. Instead, the Building Block scale <math id="mml-ieqn-2"><mo stretchy="false">(</mo><msubsup><mi>R</mi><mrow><mi>a</mi><mi>d</mi><mi>j</mi></mrow><mrow><mn>2</mn></mrow></msubsup></math> = 0.314) provides robust structural stability, filtering micro-scale noise without overfitting. A critical finding reveals the synergistic impact of physical and demographic densities. Standardized coefficients show Gross Population Density and BCR act jointly. Unstandardized baseline Multiple Linear Regression (MLR) models indicate every 10% BCR increase is associated with an approximately 0.08°C LST rise (Coef = 0.813). This research’s originality lies in its multi-dimensional approach to scale uncertainty, prioritizing ground-level breathing design over mere density control. The Building Block model’s structural stability, confirmed by consistent in-sample (0.92°C) and out-of-sample LOOCV (0.93°C) errors, provides a verified, robust toolkit for resilient urban planning.
KW  - Urban heat island; land surface temperature; building coverage ratio; multi-scale analysis; coastal cities

DO  - 10.32604/rig.2026.085793