Climate-sensitive Urban Planning through Optimization of Tree Placements
Simon Schrodi, Ferdinand Briegel, Max Argus, Andreas Christen, Thomas, Brox
TL;DR
This paper presents a neural network-based method for optimizing urban tree placements to improve thermal comfort during heatwaves, enabling large-scale and long-term climate adaptation planning.
Contribution
It introduces a novel neural network approach to simulate radiant temperatures efficiently, facilitating optimal tree placement over large areas and extended time scales.
Findings
Neural networks accurately predict radiant temperatures across various scenarios.
Optimized tree placements significantly reduce thermal discomfort during heatwaves.
The method scales effectively to large urban areas and long-term planning.
Abstract
Climate change is increasing the intensity and frequency of many extreme weather events, including heatwaves, which results in increased thermal discomfort and mortality rates. While global mitigation action is undoubtedly necessary, so is climate adaptation, e.g., through climate-sensitive urban planning. Among the most promising strategies is harnessing the benefits of urban trees in shading and cooling pedestrian-level environments. Our work investigates the challenge of optimal placement of such trees. Physical simulations can estimate the radiative and thermal impact of trees on human thermal comfort but induce high computational costs. This rules out optimization of tree placements over large areas and considering effects over longer time scales. Hence, we employ neural networks to simulate the point-wise mean radiant temperatures--a driving factor of outdoor human thermal…
Peer Reviews
Decision·Submitted to ICLR 2024
- The paper is well presented, problem statement is clearly stated, and the results are easy to follow - Use of real-world data - Claims are backed by theoretical analysis and detailed experiments.
- Lack of technical description of the algorithm. It is not clear how the hill-climbing algorithm works. Methods use domain-specific terminology that have not been adequately explained for an ML conference reader. - Application of standard ML methods, novelty is in a narrow application area - 500m x 500m is not a "large neighborhood". Prior state-of-the-art is not stated. - No details on the size of the training data, separation of train-test split details is provided. - It is unclear how the
**S1**. The work addresses a relevant, practical problem of potential societal impact. I found the analyses in 4.3 and 4.4 particularly interesting and insightful. **S2**. The paper is very well-written, clear, and easy to follow.
**W1**. In my opinion, the methodological contributions of the paper are thin, and essentially boil down to 1) estimating the aggregated radiant temperatures directly instead of individually and 2) integrating this estimation in a standard local search procedure. These are both fairly straightforward. The "theoretical analysis" in Section 3.1 is extremely tenuous and, in my opinion, should not be branded as such. **W2**. The evaluation does not include error bars and confidence intervals, which
_Originality:_ This work does not seem to be of particular originality in terms of ML methods. Hence, the manuscript might be better suited for a journal about application of ML to environmental processes. I do like the topic of research and see its importance, though, and would like to encourage the authors to submit the work to an according journal. _Quality:_ The manuscript is sound. Results of the experiments support the claims. _Clarity:_ I had some difficulties in understanding what kind
1. Unclear what data has been used. Does the CityGML data come from [this](https://www.ogc.org/standard/CityGML/) homepage and is this all simulation data, or real observations? As far as I know, ERA5 data is only available on 30m resolution; how do you get to 1m resolution? Where do your digital elevation and surface models, your land cover, wall aspects and height, and sky view factor maps come from? 2. Benchmarking traditional physical model would be highly appreciated to understand the perfo
Code & Models
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsUrban Heat Island Mitigation · Urban Green Space and Health · Land Use and Ecosystem Services