Deep Learning for Urban Microclimate Downscaling: From Coarse WRF Data to Building-Resolved PALM Simulations
Bibliographic record
Abstract
Accurate high-resolution urban microclimate modeling, including wind and temperature prediction, is essential for urban planning, occupant comfort analysis, and building energy efficiency optimization. However, traditional computational fluid dynamics (CFD) methods are computationally expensive and time-intensive for applications requiring rapid urban microclimate estimation. This work presents a novel deep learning framework that directly downscales kilometer-scale Weather Research and Forecasting (WRF) model outputs to a 10-meter level resolution 3D urban microclimate for a given geographical setting. By incorporating building geometries as model inputs, our approach captures fine-scale building-induced effects in urban wind and temperature fields, which are absent in WRF's coarse-resolution outputs.The deep learning model is trained and evaluated using urban microclimate data simulated with PALM for a realistic geographical setting in Berlin, Germany, where one week's worth of low-resolution WRF outputs serve as boundary conditions. Our proposed approach follows a two-stage training process. First, a conditional neural field (CNF) encodes the coarse WRF boundary conditions and generates a smooth, building-agnostic 3D flow field at PALM resolution. Next, a geometry-aware Fourier neural operator (FNO) refines this field by incorporating high-resolution building geometries, accurately capturing the complex interactions between airflow and urban structures. To effectively represent complex building geometries, we introduce a multi-directional distance feature (MDDF) that captures long-range spatial relationships between buildings. By producing building-resolved microclimate data from WRF outputs in near-real-time, our approach facilitates applications that are otherwise impractical with conventional CFD solvers. Despite being trained on a limited set of WRF boundary conditions, our model generalizes effectively to unseen conditions, underscoring its potential as a powerful and flexible tool for rapid urban microclimate forecasting and analysis.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".