levelIn the last level of analysis a more focused approach wasfollowed using the previous six cities so as to test the same correla-tions between noise and green in a larger scale. Specifically for thisanalysis a Geographically Weighted Regression (GWR) approachwas applied, which can better describe the geographical varia-tions between the variables instead of assuming that a single linearmodel can be fitted to the entire study area (Bristol University,2009). The parameters used in the GWR tool in ArcGIS (ESRI, 2016a)include a fixed kernel type combined with the AICc bandwidthmethod, which can identify the optimal adaptive number of neigh-bours for each case study area. The produced output refers to amultipart area composed of various 500 × 500 m-pixel blocks.