Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AutoLCZ: Towards Automatized Local Climate Zone Mapping from Rule-Based Remote Sensing

Chenying Liu, Hunsoo Song, Anamika Shreevastava, Conrad M Albrecht

TL;DR

This paper tackles scalable, interpretable Local Climate Zone mapping by introducing AutoLCZ, a framework that extracts four LiDAR-derived, rule-based parameters (BSF, ISF, PSF, and HRE) to classify LCZs within a GIS workflow. It develops data-driven thresholds derived from class-wise statistics to map LCZs in a multi-label setting, enabling LCZ classification with limited reliance on extensive manual labeling. A NYC proof-of-concept demonstrates the approach using four parameters to distinguish eight built LCZ types, showing that data-adjusted thresholds improve LCZ coverage and accuracy while also enabling quality-control insights for mislabeled labels. The work highlights AutoLCZ’s potential for large-scale, physically interpretable urban climate mapping and outlines future directions to enhance accuracy and extend modality integration.

Abstract

Local climate zones (LCZs) established a standard classification system to categorize the landscape universe for improved urban climate studies. Existing LCZ mapping is guided by human interaction with geographic information systems (GIS) or modelled from remote sensing (RS) data. GIS-based methods do not scale to large areas. However, RS-based methods leverage machine learning techniques to automatize LCZ classification from RS. Yet, RS-based methods require huge amounts of manual labels for training. We propose a novel LCZ mapping framework, termed AutoLCZ, to extract the LCZ classification features from high-resolution RS modalities. We study the definition of numerical rules designed to mimic the LCZ definitions. Those rules model geometric and surface cover properties from LiDAR data. Correspondingly, we enable LCZ classification from RS data in a GIS-based scheme. The proposed AutoLCZ method has potential to reduce the human labor to acquire accurate metadata. At the same time, AutoLCZ sheds light on the physical interpretability of RS-based methods. In a proof-of-concept for New York City (NYC) we leverage airborne LiDAR surveys to model 4 LCZ features to distinguish 10 LCZ types. The results indicate the potential of AutoLCZ as promising avenue for large-scale LCZ mapping from RS data.

AutoLCZ: Towards Automatized Local Climate Zone Mapping from Rule-Based Remote Sensing

TL;DR

This paper tackles scalable, interpretable Local Climate Zone mapping by introducing AutoLCZ, a framework that extracts four LiDAR-derived, rule-based parameters (BSF, ISF, PSF, and HRE) to classify LCZs within a GIS workflow. It develops data-driven thresholds derived from class-wise statistics to map LCZs in a multi-label setting, enabling LCZ classification with limited reliance on extensive manual labeling. A NYC proof-of-concept demonstrates the approach using four parameters to distinguish eight built LCZ types, showing that data-adjusted thresholds improve LCZ coverage and accuracy while also enabling quality-control insights for mislabeled labels. The work highlights AutoLCZ’s potential for large-scale, physically interpretable urban climate mapping and outlines future directions to enhance accuracy and extend modality integration.

Abstract

Local climate zones (LCZs) established a standard classification system to categorize the landscape universe for improved urban climate studies. Existing LCZ mapping is guided by human interaction with geographic information systems (GIS) or modelled from remote sensing (RS) data. GIS-based methods do not scale to large areas. However, RS-based methods leverage machine learning techniques to automatize LCZ classification from RS. Yet, RS-based methods require huge amounts of manual labels for training. We propose a novel LCZ mapping framework, termed AutoLCZ, to extract the LCZ classification features from high-resolution RS modalities. We study the definition of numerical rules designed to mimic the LCZ definitions. Those rules model geometric and surface cover properties from LiDAR data. Correspondingly, we enable LCZ classification from RS data in a GIS-based scheme. The proposed AutoLCZ method has potential to reduce the human labor to acquire accurate metadata. At the same time, AutoLCZ sheds light on the physical interpretability of RS-based methods. In a proof-of-concept for New York City (NYC) we leverage airborne LiDAR surveys to model 4 LCZ features to distinguish 10 LCZ types. The results indicate the potential of AutoLCZ as promising avenue for large-scale LCZ mapping from RS data.
Paper Structure (10 sections, 2 equations, 2 figures, 3 tables)

This paper contains 10 sections, 2 equations, 2 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Study area & data
  3. Methodology
  4. LCZ Parameter Modeling
  5. LCZ Classification
  6. Experiments
  7. Estimated LCZ Parameter Thresholds
  8. Rule-Based LCZ Parameters
  9. LCZ Misclassified vs. Mislabeled
  10. Conclusions

Figures (2)

  • Figure 1: Overview on the data of our study: (a) sampling areas in NYC boroughs Manhattan, Queens, Bronx, and Brooklyn, where each point represents a 0.64$^2$ km$^2$ squared region. Illustrations of a built type compact low-rise (LCZ 3) from (b) the optical view, (c) the 2D mean elevation statistics rasterized from LiDAR data, (d) the ground truth land cover mask, and (e) the noisy land cover mask derived from LiDAR statistics (cf. albrecht_autogeolabel_2021).
  • Figure 2: An example of a potentially incorrectly annotated LCZ 4 patch with corresponding (d) OpenStreetMap data and (e) Google street view photo. The red arrow in (d) points towards to the direction from which the street view photo was taken.