ON ACCELERATION OF THERMAL SIMULATION OF URBAN SCENES WITH THE APPLICATION OF AN EVOLUTIONARY ALGORITHM TO TREE PLANTING STRATEGIES

Tree planting is one of the most popular in urban morphology measures for urban heat island reduction since, at a relatively low monetary cost and with a marginal alteration of the scene, trees provide shadows and neutralize harmful gases. Findings for their optimal distribution within the urban sce...

Full description

Saved in:
Bibliographic Details
Published inISPRS annals of the photogrammetry, remote sensing and spatial information sciences Vol. X-1/W1-2023; pp. 9 - 16
Main Authors Bulatov, D., Hecht, M., Kottler, B., Mispelhorn, J., Strauss, E.
Format Journal Article
LanguageEnglish
Published Gottingen Copernicus GmbH 01.01.2023
Copernicus Publications
Subjects
Algorithms
Computer graphics
Digital twins
Environmental studies
Evolutionary algorithms
Genetic algorithms
Rendering
Thermal simulation
Tree planting
Urban heat islands
Online AccessGet full text

Cover

More Information
Summary:Tree planting is one of the most popular in urban morphology measures for urban heat island reduction since, at a relatively low monetary cost and with a marginal alteration of the scene, trees provide shadows and neutralize harmful gases. Findings for their optimal distribution within the urban scene exist in numerous environmental studies. However, merely the digital twin of the scene possesses the capability to analyze further developments of the scene, such as changes in dominant wind directions. Today’s extensive computational resources allow for thermal simulations of digital twins on multiple (not-yet-existing) urban scene designs, aiming at minimizing the average or peak temperatures. From the point of view of computer graphics, this paper proposes four tools to accelerate an evolutionary algorithm for tree planting strategies. Using GPU arrays for rendering, pre-rendering default scenes, and pre-filtering trees in the early morning and late evening hours helps accelerating the rendering process. Computation of fitness function on different computers allows a further acceleration of the evolutionary algorithm. The total acceleration factor of a scene using computational set-up exceeds 218, thus demonstrating the enormous potential the evolutionary algorithm may bring about in future investigations.
ISSN:2194-9050
2194-9042
2194-9050
DOI:10.5194/isprs-annals-X-1-W1-2023-9-2023