Spatial Modeling for Building Design Evaluation: from Visual Landscape Quality Assessment to Devaluation Risk Estimation

Zoning reform is a crucial tool for cities to adapt to contemporary challenges. However, its implementation remains challenging. Property owners, with a vested interest in the value of their neighborhoods, are sensitive to local developments and the potential unforeseen effects on environmental amenities. Where complete information exists, environmental amenities and risks are priced into real estate valuations. Yet, there remains a lack of forward-looking micro-scale environmental data. Moreover, methods to incorporate such information into urban and building design evaluation are limited to the macro-scale: e.g. climate change risk. These hinder the markets' ability to effectively price discriminate, especially concerning uninsured local risks like zoning reform. Addressing this gap, this thesis leverages advancements in spatial modeling and geographical artificial intelligence (GeoAI) to estimate the financial impact of such risks, with a focus on the devaluation risk of visual impact from urban densification.This thesis introduces spatial modeling for building design evaluation as four parts: (i) Performance Simulation, (ii) Design Evaluation, (iii) Environmental Valuation, and (iv) Design Impact. The concept of design performance, its economic evaluation, and its exposure and sensitivity are introduced by relating building performance simulations and real estate economics.This work explores the impact of environmental risks on real estate valuation and proposes the concept of local area devaluation risk estimation. It focuses on visual impact resulting from nearby land-use changes as the variable of interest due to (1) the influence window views have on property valuations and on public opposition to densification and (2) the lack of methods to measure building-level visual quality in a comprehensive manner. It presents methods for the (i) 3D-CAD simulation of viewpoint visual shares and the (ii) statistical analysis of Visual Capital, a unique approach that estimates building level visual landscape quality by modeling income-sorting. Further, it introduces the (iii) hedonic pricing of Visual Capital and its application in an (iv) integrated impact analysis of computationally generated urban scenarios through Architectural Design Appraisal. Design Appraisal forecasts prices of procedurally generated building designs, using the learned estimates of financial preference of design performance. Applied within a regional simulation, it introduces a property-level environmental impact assessment to study local area devaluation risk.This work represents the first-time that financial valuation is integrated within building design evaluation. The main results illustrate the potential of large-scale 3D data and GeoAI to (1) capture difficult to assess urban amenities (e.g. the view) and risks (e.g. obstruction), and to (2) inform urban design and land-use by incorporating market information into design simulations. This thesis concludes with a discussion of how these new concepts facilitate preference-driven generative design optimization and site selection.