Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

GSR4B: Biomass Map Super-Resolution with Sentinel-1/2 Guidance

Kaan Karaman, Yuchang Jiang, Damien Robert, Vivien Sainte Fare Garnot, Maria João Santos, Jan Dirk Wegner

Abstract

Accurate Above-Ground Biomass (AGB) mapping at both large scale and high spatio-temporal resolution is essential for applications ranging from climate modeling to biodiversity assessment, and sustainable supply chain monitoring. At present, fine-grained AGB mapping relies on costly airborne laser scanning acquisition campaigns usually limited to regional scales. Initiatives such as the ESA CCI map attempt to generate global biomass products from diverse spaceborne sensors but at a coarser resolution. To enable global, high-resolution (HR) mapping, several works propose to regress AGB from HR satellite observations such as ESA Sentinel-1/2 images. We propose a novel way to address HR AGB estimation, by leveraging both HR satellite observations and existing low-resolution (LR) biomass products. We cast this problem as Guided Super-Resolution (GSR), aiming at upsampling LR biomass maps (sources) from $100$ to $10$ m resolution, using auxiliary HR co-registered satellite images (guides). We compare super-resolving AGB maps with and without guidance, against direct regression from satellite images, on the public BioMassters dataset. We observe that Multi-Scale Guidance (MSG) outperforms direct regression both for regression ($-780$ t/ha RMSE) and perception ($+2.0$ dB PSNR) metrics, and better captures high-biomass values, without significant computational overhead. Interestingly, unlike the RGB+Depth setting they were originally designed for, our best-performing AGB GSR approaches are those that most preserve the guide image texture. Our results make a strong case for adopting the GSR framework for accurate HR biomass mapping at scale. Our code and model weights are made publicly available (https://github.com/kaankaramanofficial/GSR4B).

GSR4B: Biomass Map Super-Resolution with Sentinel-1/2 Guidance

Abstract

Accurate Above-Ground Biomass (AGB) mapping at both large scale and high spatio-temporal resolution is essential for applications ranging from climate modeling to biodiversity assessment, and sustainable supply chain monitoring. At present, fine-grained AGB mapping relies on costly airborne laser scanning acquisition campaigns usually limited to regional scales. Initiatives such as the ESA CCI map attempt to generate global biomass products from diverse spaceborne sensors but at a coarser resolution. To enable global, high-resolution (HR) mapping, several works propose to regress AGB from HR satellite observations such as ESA Sentinel-1/2 images. We propose a novel way to address HR AGB estimation, by leveraging both HR satellite observations and existing low-resolution (LR) biomass products. We cast this problem as Guided Super-Resolution (GSR), aiming at upsampling LR biomass maps (sources) from to m resolution, using auxiliary HR co-registered satellite images (guides). We compare super-resolving AGB maps with and without guidance, against direct regression from satellite images, on the public BioMassters dataset. We observe that Multi-Scale Guidance (MSG) outperforms direct regression both for regression ( t/ha RMSE) and perception ( dB PSNR) metrics, and better captures high-biomass values, without significant computational overhead. Interestingly, unlike the RGB+Depth setting they were originally designed for, our best-performing AGB GSR approaches are those that most preserve the guide image texture. Our results make a strong case for adopting the GSR framework for accurate HR biomass mapping at scale. Our code and model weights are made publicly available (https://github.com/kaankaramanofficial/GSR4B).

Paper Structure

This paper contains 23 sections, 3 equations, 5 figures, 2 tables.

Table of Contents

  1. Biomass Estimation
  2. Biomass datasets.
  3. Biomass estimation from remote sensing data.
  4. Guided Super-Resolution
  5. GDSR approaches.
  6. Super-resolution methods.
  7. JBU.
  8. MSG.
  9. P2P.
  10. FDSR.
  11. Dataset description.
  12. Downsampling for source maps.
  13. Overall performance.
  14. Better performance at high AGB.
  15. Comparison between GSR methods.
  16. ...and 8 more sections

Figures (5)

  • Figure 1: Given a LR AGB map and co-registered HR satellite imagery, our method predicts the upsampled HR biomass map. Increasing AGB values are represented from white to green.
  • Figure 2: The box plot shows the errors of the predictions versus the ground truth AGB values (without outliers for simplicity). For each box, we randomly select $10~000$ pixels from the test set. px$~=100~\text{m}^2$, Error $=$ Model Prediction $-$ Ground Truth AGB
  • Figure 3: The first two rows show the predicted AGB maps from various methods alongside the ground truth high-resolution maps of a random test sample. Lower to higher AGB values range from red to blue. The last two rows show the residuals between the reference and predicted maps. Joint Bilateral Upsampling (JBU), Pixel-to-Pixel Transform (P2P), Fast Depth Map Super-Resolution (FDSR) and Deep Multi-Scale Guidance (MSG) models are originally the guided depth super-resolution methods, tested on the biomass dataset.
  • Figure 4: The frequency response of the outputs on the test sample in Figure \ref{['fig:PredictionsForFixedSamples']}.
  • Figure 5: The histogram of the frequency responses of deep-learning-based guided super-resolution models.