Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Joint Super-Resolution and Segmentation for 1-m Impervious Surface Area Mapping in China's Yangtze River Economic Belt

Jie Deng, Danfeng Hong, Chenyu Li, Naoto Yokoya

TL;DR

This work addresses the challenge of obtaining fine-scale ISA maps over large regions by combining progressive super-resolution with semantic segmentation. The authors introduce JointSeg, pairing Prog-ESRGAN SR with Mask2Former segmentation to produce 1 m ISA maps directly from 10 m Sentinel-2 data, enabling large-scale, temporally consistent ISA mapping for the 2.4 million km^2 Yangtze River Economic Belt. The ISA-1 product demonstrates strong performance (OA ≈ 93.7%, ISA F1 ≈ 0.76) and reveals nuanced urban-rural and topographic patterns, outperforming existing 10 m and 30 m datasets, with biennial analyses from 2017–2023 showing region-specific growth trajectories. This approach democratizes high-resolution land-cover information by leveraging freely available imagery, and sets a foundation for near-real-time urban dynamics monitoring and policy-relevant planning, while also highlighting avenues for end-to-end training and domain adaptation in future work.

Abstract

We propose a novel joint framework by integrating super-resolution and segmentation, called JointSeg, which enables the generation of 1-meter ISA maps directly from freely available Sentinel-2 imagery. JointSeg was trained on multimodal cross-resolution inputs, offering a scalable and affordable alternative to traditional approaches. This synergistic design enables gradual resolution enhancement from 10m to 1m while preserving fine-grained spatial textures, and ensures high classification fidelity through effective cross-scale feature fusion. This method has been successfully applied to the Yangtze River Economic Belt (YREB), a region characterized by complex urban-rural patterns and diverse topography. As a result, a comprehensive ISA mapping product for 2021, referred to as ISA-1, was generated, covering an area of over 2.2 million square kilometers. Quantitative comparisons against the 10m ESA WorldCover and other benchmark products reveal that ISA-1 achieves an F1-score of 85.71%, outperforming bilinear-interpolation-based segmentation by 9.5%, and surpassing other ISA datasets by 21.43%-61.07%. In densely urbanized areas (e.g., Suzhou, Nanjing), ISA-1 reduces ISA overestimation through improved discrimination of green spaces and water bodies. Conversely, in mountainous regions (e.g., Ganzi, Zhaotong), it identifies significantly more ISA due to its enhanced ability to detect fragmented anthropogenic features such as rural roads and sparse settlements, demonstrating its robustness across diverse landscapes. Moreover, we present biennial ISA maps from 2017 to 2023, capturing spatiotemporal urbanization dynamics across representative cities. The results highlight distinct regional growth patterns: rapid expansion in upstream cities, moderate growth in midstream regions, and saturation in downstream metropolitan areas.

Joint Super-Resolution and Segmentation for 1-m Impervious Surface Area Mapping in China's Yangtze River Economic Belt

TL;DR

This work addresses the challenge of obtaining fine-scale ISA maps over large regions by combining progressive super-resolution with semantic segmentation. The authors introduce JointSeg, pairing Prog-ESRGAN SR with Mask2Former segmentation to produce 1 m ISA maps directly from 10 m Sentinel-2 data, enabling large-scale, temporally consistent ISA mapping for the 2.4 million km^2 Yangtze River Economic Belt. The ISA-1 product demonstrates strong performance (OA ≈ 93.7%, ISA F1 ≈ 0.76) and reveals nuanced urban-rural and topographic patterns, outperforming existing 10 m and 30 m datasets, with biennial analyses from 2017–2023 showing region-specific growth trajectories. This approach democratizes high-resolution land-cover information by leveraging freely available imagery, and sets a foundation for near-real-time urban dynamics monitoring and policy-relevant planning, while also highlighting avenues for end-to-end training and domain adaptation in future work.

Abstract

We propose a novel joint framework by integrating super-resolution and segmentation, called JointSeg, which enables the generation of 1-meter ISA maps directly from freely available Sentinel-2 imagery. JointSeg was trained on multimodal cross-resolution inputs, offering a scalable and affordable alternative to traditional approaches. This synergistic design enables gradual resolution enhancement from 10m to 1m while preserving fine-grained spatial textures, and ensures high classification fidelity through effective cross-scale feature fusion. This method has been successfully applied to the Yangtze River Economic Belt (YREB), a region characterized by complex urban-rural patterns and diverse topography. As a result, a comprehensive ISA mapping product for 2021, referred to as ISA-1, was generated, covering an area of over 2.2 million square kilometers. Quantitative comparisons against the 10m ESA WorldCover and other benchmark products reveal that ISA-1 achieves an F1-score of 85.71%, outperforming bilinear-interpolation-based segmentation by 9.5%, and surpassing other ISA datasets by 21.43%-61.07%. In densely urbanized areas (e.g., Suzhou, Nanjing), ISA-1 reduces ISA overestimation through improved discrimination of green spaces and water bodies. Conversely, in mountainous regions (e.g., Ganzi, Zhaotong), it identifies significantly more ISA due to its enhanced ability to detect fragmented anthropogenic features such as rural roads and sparse settlements, demonstrating its robustness across diverse landscapes. Moreover, we present biennial ISA maps from 2017 to 2023, capturing spatiotemporal urbanization dynamics across representative cities. The results highlight distinct regional growth patterns: rapid expansion in upstream cities, moderate growth in midstream regions, and saturation in downstream metropolitan areas.
Paper Structure (20 sections, 10 figures, 3 tables)

This paper contains 20 sections, 10 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Dataset and Methods
  3. Study area
  4. ISA Super-resolution segmentation dataset (ISASeg)
  5. Modelling
  6. Super-Resolution Model
  7. Semantic Segmentation Model
  8. Training Strategy
  9. Results
  10. Super-resolution results
  11. Semantic Segmentation Result
  12. Large-Scale Regional Comparison
  13. Analysis of Biennial ISA Products from 2017 to 2023
  14. Discussion
  15. Performance of Different Super-resolution Models
  16. ...and 5 more sections

Figures (10)

  • Figure 1: (a)The Yangtze River Economic Belt (YREB, shown in blue) is one of the most urbanized and economically dynamic regions in the world. Between 2010 and 2020, its urban population grew by 109 million, and its GDP increased by 163.8% China_2024YREB. (b)The YREB covers 27.7% of China’s land, hosts 44.2% of its population, contributes 47.7% of GDP, and contains 32.1% of arable land and 63.3% of paddy fields China_2024YREB. (c)The elevation area covers an elevation gradient from 31 meters to 7,083 meters and a total area of over 720,000 square kilometers.
  • Figure 2: Distribute sample points within the study area while ensuring that the minimum distance between any two points is 3 km to prevent overlap, and provide statistics for the ISASeg dataset.
  • Figure 3: The workflow for generating high-resolution impervious surface area (ISA) labels begins with extracting land use/land cover (LULC) categories and road networks from OpenStreetMap. This is followed by incorporating building labels, semi-automated annotation software, and high-resolution RGB imagery to refine the ISA labels further.
  • Figure 4: Workflow for generating the Artificial impervious surface area (ISA-1, 1m) using Sentinel 2 (10m) satellite images and super-resolution segmentation pipeline.
  • Figure 5: The architecture of the super-resolution segmentation (JointSeg) model.
  • ...and 5 more figures