Land Surface Temperature Retrieval From GF5-01A Based on Split- Window Algorithm

Yu Liu, Yongxin Wang, Guanghui Wang2024 · DOI

TL;DR

This paper introduces a new split-window algorithm for retrieving Land Surface Temperature (LST) from China's GF5-01A satellite, which boasts a wider swath and more thermal bands than previous sensors. The algorithm was developed using atmospheric and spectral libraries and validated against Landsat 8/9 LST products and ground-based SURFRAD measurements, showing promising accuracy for global climate monitoring.

What Problem Does This Paper Address?

Accurate and wide-area monitoring of Land Surface Temperature (LST) is crucial for understanding Earth's energy balance, climate change, and ecological processes. Traditional ground-based measurements are labor-intensive and spatially limited. Satellite-based thermal infrared remote sensing offers a solution for rapid, large-scale LST acquisition. However, atmospheric interference and sensor limitations can affect retrieval accuracy. This paper addresses the need for a robust LST retrieval algorithm for the newly launched GF5-01A satellite, which offers enhanced capabilities (100m resolution, 1500km swath) but requires a tailored algorithm to leverage its potential.

Methods

The core of this research is the development and validation of a split-window algorithm for LST retrieval from GF5-01A. This is a well-established multi-channel technique that exploits the differential atmospheric absorption in two adjacent thermal infrared bands (around 11 and 12 µm) to correct for atmospheric effects.

The authors followed these key steps:

  • Algorithm Construction:
    • They used the GF5-01A's Wide Swath Thermal Infrared (WTI) imager's spectral response functions.
  • Atmospheric profiles from the TIGR2000 dataset and spectral libraries from ASTER were combined.
  • The MODTRAN 5.2 radiative transfer model was employed to simulate thermal infrared radiance under clear-sky conditions.
  • Based on these simulations, a split-window algorithm of the form $T_s = a_0 + a_1T_i + a_2(T_i - T_j) + a_3(T_i - T_j)^2$ was constructed, where $T_s$ is the retrieved LST, $T_i$ and $T_j$ are the brightness temperatures from the two GF5-01A thermal bands, and $a_0$ to $a_3$ are the algorithm coefficients.
  • Validation:
    • Cross-validation: GF5-01A LST products were compared with existing, validated LST products from Landsat 8/9. This involved registering the datasets and selecting representative validation points over water, vegetation, and sand surfaces.
  • Temperature-based method: GF5-01A LST retrievals were compared with in-situ measurements from the SURFRAD network. This involved converting SURFRAD's longwave radiation measurements to LST using the Stefan-Boltzmann law and surface broadband emissivity derived from ASTER GED products.

Datasets

  • GF5-01A WTI Imager Data: The primary sensor data used for LST retrieval.
  • TIGR2000 Atmospheric Profiles: A dataset of atmospheric profiles used for simulating atmospheric conditions.
  • ASTER Spectral Library: Used to obtain spectral response functions for surface emissivity.
  • MODTRAN 5.2: A radiative transfer model used for data simulation.
  • Landsat 8/9 LST Products: Used for cross-validation of the GF5-01A LST retrievals.
  • SURFRAD Network Data: In-situ ground measurements of surface radiation budget used for direct validation of GF5-01A LST retrievals.

Key Insights and Ideas

  • GF5-01A's Potential: The GF5-01A satellite, with its 100m resolution and 1500km swath, offers a significant advancement for global LST monitoring compared to sensors like Landsat. The increased number of thermal bands (4 vs. 2 for Landsat TIRS) also provides more flexibility for algorithm development.
  • Split-Window Algorithm Efficacy: The study demonstrates that a split-window algorithm, a relatively simple yet effective method, can be successfully adapted for GF5-01A data, achieving reasonable accuracy.
  • Land Cover Influence on Accuracy: The validation results highlight that LST retrieval accuracy can vary with land cover type. Water bodies showed the highest accuracy, while sand surfaces were the most challenging, likely due to their unevenness and susceptibility to directional radiation effects.
  • Day vs. Night Performance: The paper suggests that validation results for nighttime data might be better than daytime data, possibly due to the more stable thermal properties of heterogeneous surfaces at night compared to the influence of solar irradiation and shadow areas during the day.

Key Results

  • Cross-validation with Landsat 8/9:
    • Overall RMSE: 1.72 K
    • Overall Bias: 0.41 K
    • Water bodies: RMSE 1.27 K, Bias 1.12 K
    • Sand surfaces: RMSE 2.24 K, Bias -2.08 K
  • Validation with SURFRAD:
    • Overall RMSE: 2.11 K
    • Overall Bias: 0.68 K
    • Individual site RMSEs ranged from 0.68 K to 2.64 K.

These results indicate that the developed GF5-01A split-window algorithm meets the requirements for thermal infrared remote sensing monitoring and shows considerable potential.

Limitations and Future Work

  • Limited SURFRAD Data: The authors acknowledge that the validation with SURFRAD data was based on a small sample size (9 pairs of data, covering 4 sites).
  • Atmospheric Conditions: The current algorithm was developed and validated primarily under clear-sky conditions.
  • Surface Heterogeneity: The accuracy for certain land cover types, like sand, was lower due to surface roughness and directional radiation effects.
  • Future Work:
    • Conduct more in-depth accuracy evaluations.
  • Expand validation efforts with a larger amount of measured data.
  • Perform validation under a wider range of atmospheric conditions and land-cover types.
  • Explore the potential of GF5-01A's multiple thermal bands for more advanced LST retrieval algorithms.

Summary updated

Abstract

Land Surface temperature (LST) is a core parameter in the energy exchange between the surface and the atmosphere, and the use of thermal infrared remote sensing can realize the wide-range, fast, and accurate acquisition of surface temperature. GF5-01A is an important part and the final satellite of the major special project on high-resolution Earth observation system which is equipped with a wide-area thermal infrared imager with a resolution of 100 meters and a width of 1,500 kilometers. In this paper, based on the GF5-01A WTI spectral response function, combined with the TIGR2000 atmospheric profiles data and the ASTER spectral library, the data simulation was carried out by using the atmospheric radiative transfer model MODTRAN 5.2, and then constructed the split-window algorithm. Then, the method proposed in this paper was validated and evaluated using Landsat 8/9 temperature products and measured surface temperature data from SURFRAD sites acquired on the same day. The results show that the RMSE between the GF5-01A retrieved LST and the Landsat8/9 retrieved LST is between 1.27–2.24K, and the Bias is between −2.08–1.12K. The RMSE is between 0.68–2.64K and the bias is between −0.68–1.49K compared to the measured surface temperature. The split-window algorithm of GF5-01A proposed in this paper can meet the requirements of thermal infrared remote sensing monitoring and has enormous potential and value.
BibTeX
@inproceedings{liu2024land,
  author = {Yu Liu and Yongxin Wang and Guanghui Wang},
  title = {{Land Surface Temperature Retrieval From GF5-01A Based on Split- Window Algorithm}},
  booktitle = {˜The},
  year = {2024},
  doi = {10.5194/isprs-archives-xlviii-1-2024-431-2024},
}
Figure