Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Downscaling land surface temperature data using edge detection and block-diagonal Gaussian process regression

Sanjit Dandapanthula, Margaret Johnson, Madeleine Pascolini-Campbell, Glynn Hulley, Mikael Kuusela

TL;DR

The paper tackles high-resolution downscaling of land surface temperature (LST) by leveraging Landsat-derived agricultural field structure to inform a block-diagonal Gaussian process (BDGP) that captures field-structured spatial variation and addresses the change of support between ECOSTRESS and Landsat. The approach combines edge-detected field partitions, Fourier-based mean modeling, region-wise maximum likelihood parameter estimation, and BDGP kriging to reconstruct high-resolution LST with uncertainty quantification. Key contributions include a practical BDGP framework for data fusion across sensors with different resolutions, a principled method for mean and variance estimation under blur, and scalable, region-wise inference. The results demonstrate that the method can deblur ECOSTRESS LST images and produce reliable high-resolution estimates with quantified uncertainty, highlighting potential applications in agriculture, urban planning, and climate studies. The work also outlines limitations and avenues for extension, including temporal dynamics, alternative segmentation strategies, and reconstruction of road networks.

Abstract

Accurate and high-resolution estimation of land surface temperature (LST) is crucial in estimating evapotranspiration, a measure of plant water use and a central quantity in agricultural applications. In this work, we develop a novel statistical method for downscaling LST data obtained from NASA's ECOSTRESS mission, using high-resolution data from the Landsat 8 mission as a proxy for modeling agricultural field structure. Using the Landsat data, we identify the boundaries of agricultural fields through edge detection techniques, allowing us to capture the inherent block structure present in the spatial domain. We propose a block-diagonal Gaussian process (BDGP) model that captures the spatial structure of the agricultural fields, leverages independence of LST across fields for computational tractability, and accounts for the change of support present in ECOSTRESS observations. We use the resulting BDGP model to perform Gaussian process regression and obtain high-resolution estimates of LST from ECOSTRESS data, along with uncertainty quantification. Our results demonstrate the practicality of the proposed method in producing reliable high-resolution LST estimates, with potential applications in agriculture, urban planning, and climate studies.

Downscaling land surface temperature data using edge detection and block-diagonal Gaussian process regression

TL;DR

The paper tackles high-resolution downscaling of land surface temperature (LST) by leveraging Landsat-derived agricultural field structure to inform a block-diagonal Gaussian process (BDGP) that captures field-structured spatial variation and addresses the change of support between ECOSTRESS and Landsat. The approach combines edge-detected field partitions, Fourier-based mean modeling, region-wise maximum likelihood parameter estimation, and BDGP kriging to reconstruct high-resolution LST with uncertainty quantification. Key contributions include a practical BDGP framework for data fusion across sensors with different resolutions, a principled method for mean and variance estimation under blur, and scalable, region-wise inference. The results demonstrate that the method can deblur ECOSTRESS LST images and produce reliable high-resolution estimates with quantified uncertainty, highlighting potential applications in agriculture, urban planning, and climate studies. The work also outlines limitations and avenues for extension, including temporal dynamics, alternative segmentation strategies, and reconstruction of road networks.

Abstract

Accurate and high-resolution estimation of land surface temperature (LST) is crucial in estimating evapotranspiration, a measure of plant water use and a central quantity in agricultural applications. In this work, we develop a novel statistical method for downscaling LST data obtained from NASA's ECOSTRESS mission, using high-resolution data from the Landsat 8 mission as a proxy for modeling agricultural field structure. Using the Landsat data, we identify the boundaries of agricultural fields through edge detection techniques, allowing us to capture the inherent block structure present in the spatial domain. We propose a block-diagonal Gaussian process (BDGP) model that captures the spatial structure of the agricultural fields, leverages independence of LST across fields for computational tractability, and accounts for the change of support present in ECOSTRESS observations. We use the resulting BDGP model to perform Gaussian process regression and obtain high-resolution estimates of LST from ECOSTRESS data, along with uncertainty quantification. Our results demonstrate the practicality of the proposed method in producing reliable high-resolution LST estimates, with potential applications in agriculture, urban planning, and climate studies.
Paper Structure (12 sections, 15 equations, 10 figures)

This paper contains 12 sections, 15 equations, 10 figures.

Table of Contents

  1. Introduction
  2. Methods
  3. Edge detection
  4. Fourier regression
  5. Parameter estimation
  6. Kriging and uncertainty quantification
  7. Results
  8. Discussion
  9. Estimating the ECOSTRESS variance parameter
  10. Additional figures
  11. Verification of methods
  12. Refining the oversegmented image

Figures (10)

  • Figure 1: Comparison of resolution between ECOSTRESS and Landsat missions over the Salton Sea region in southern California.
  • Figure 2: Overview of the proposed downscaling pipeline. Overview of the proposed downscaling pipeline. The Landsat data is shown in blue, the ECOSTRESS data is shown in orange, and the output of the algorithm is shown in purple.
  • Figure 3: Side-by-side comparison of (a) the averaged Landsat image, (b) the oversegmented image from SAM, and (c) the final regions after post-processing.
  • Figure 4: Side-by-side comparison of (a) the original ECOSTRESS image, (b) the ECOSTRESS residuals after subtracting the fitted mean function, (c) the original Landsat image, and (d) the Landsat residuals after subtracting the fitted mean function.
  • Figure 5: Fitted parameters for each region: (a) Landsat per-region variance, (b) per-region length scales (shared between Landsat and ECOSTRESS), and (c) ECOSTRESS per-region variance.
  • ...and 5 more figures