The global leaf chlorophyll content dataset over 2003–2012 and 2018–2020 derived from MERIS/OLCI satellite data (GLCC): algorithm and validation

Xiaojin Qian, Liangyun Liu, Xidong Chen, Xiao Zhang, Siyuan Chen, Qi Sun2022 · DOI

TL;DR

This paper introduces a new global Leaf Chlorophyll Content (LCC) dataset (GLCC) spanning nearly 20 years (2003-2012 and 2018-2020) by combining data from the MERIS and Sentinel-3 OLCI satellites. This dataset, generated using an improved physically-based radiative transfer modeling approach, offers a valuable resource for monitoring vegetation health and understanding global carbon cycles.

What Problem Does This Paper Address?

Quantifying global photosynthesis is crucial for understanding Earth's carbon cycle, and leaf chlorophyll content (LCC) is a key indicator of photosynthetic capacity. However, accurately measuring LCC globally is challenging due to the limitations of ground-based observations (spatially and temporally sparse) and the complexity of satellite data. Previous efforts, like the MERIS-based LCC dataset, were limited by the lifespan of the MERIS sensor. This paper addresses the need for a continuous, long-term global LCC dataset by leveraging both historical MERIS data and newer Sentinel-3 OLCI data, extending the temporal coverage and providing a more robust record for climate and ecosystem studies.

Methods

The researchers developed an improved look-up-table (LUT) based inversion approach using physically-based radiative transfer models. They generated LUTs by simulating canopy reflectance using the PROSPECT-D (leaf optical properties) and 4SAIL (canopy reflectance) models. For different plant functional types (PFTs), specific LUTs were created: one for non-woody vegetation (grasslands, crops) with 25 sub-LUTs accounting for canopy structure and soil background, and four for woody vegetation (forests, shrubs) with three sub-LUTs each, considering different crown heights. The final LCC retrieval for each PFT was derived by averaging the results from the relevant sub-LUTs. This approach was validated using synthetic datasets and then applied to MERIS and OLCI surface reflectance data to generate the global GLCC dataset.

Datasets

  • MERIS Surface Reflectance (SR) product: Used for the 2003-2012 period.
  • Sentinel-3 SY_2_SYN product: Used for the 2018-2020 period, providing surface reflectance from OLCI.
  • MODIS Land Cover Type (MCD12Q1) Version 6 product: Used to classify land cover into relevant PFTs.
  • Croft MERIS LCC dataset: Used for comparison and validation.
  • 161 field measurements: Collected across six PFTs, used for direct validation of the GLCC dataset.

Key Insights and Ideas

  • Bridging the MERIS-OLCI Gap: This work successfully combines data from the legacy MERIS sensor and the newer Sentinel-3 OLCI sensor to create a seamless, long-term LCC dataset. This is crucial for maintaining continuity in Earth observation records.
  • Improved LUT Approach: The refined LUT method, incorporating more detailed canopy structure and soil background variations, demonstrates enhanced accuracy in LCC retrieval compared to previous approaches.
  • PFT-Specific Modeling: Tailoring LUTs for different PFTs (woody vs. non-woody, and further subdivisions) acknowledges the significant structural and optical differences in vegetation, leading to more accurate retrievals.
  • Global Scale Validation: The validation against a substantial number of field measurements across diverse PFTs provides confidence in the dataset's reliability for global applications.

Key Results

  • Synthetic Dataset Validation: The LUT algorithm achieved high performance on synthetic data, with R² values above 0.79 and RMSE below 10.5 µg cm⁻².
  • Field Measurement Validation: The GLCC dataset showed good agreement with 161 field measurements, yielding an overall R² of 0.41 and an RMSE of 8.94 µg cm⁻². Individual PFTs showed varying levels of accuracy, with EBF performing best.
  • Consistency with Existing Data: The GLCC dataset demonstrated acceptable consistency with the existing Croft MERIS LCC dataset.
  • Temporal Coverage: The dataset spans nearly two decades (2003-2012 and 2018-2020), providing a valuable long-term record.
  • Spatial Resolution: The dataset is provided at a 500 m spatial resolution.

Limitations and Future Work

  • Field Data Limitations: The authors acknowledge limitations in the validation due to the lack of extensive, seasonally varying LCC field measurements across all PFTs, particularly for SHR, EBF, and GRA. They suggest that upscaling ground-measured data using high-resolution imagery could improve future validation.
  • Pixel Heterogeneity: The 300 m/500 m resolution of the satellite data can be affected by pixel heterogeneity, potentially introducing spatial heterogeneity in the derived LCC products. Further evaluation of this effect is recommended.
  • Model Assumptions: The fixed biochemical and structural parameters used in the LUT generation, while PFT-specific, may still influence simulation accuracy. The trade-off between model universality, computational efficiency, and inversion accuracy is noted.
  • Future Work: The paper highlights the dataset's utility for analyzing vegetation physiological dynamics and terrestrial carbon cycle modeling, suggesting its application in these areas. Further investigation into the impact of pixel heterogeneity on LUT inversion is also suggested.

Summary updated

Abstract

Leaf chlorophyll content (LCC), a prominent plant physiological trait and a proxy for leaf photosynthetic capacity, plays a crucial role in the monitoring of agriculture and carbon cycle modeling. In this study, global 500 m LCC weekly dataset (GLCC) for the period 2003–2012 to 2018–2020 were produced from ENVISAT MERIS and Sentinel-3 OLCI satellite data using a physically-based radiative transfer modeling approach. Firstly, five look-up-tables (LUTs) were generated using PROSAIL-D and PROSPECT-D+4-Scale models for woody and non-woody plants, respectively. For the four LUTs applicable to woody plants, each LUT contains three sub-LUTs corresponding to three types of crown height. For the one LUT applicable to non-woody vegetation type, it includes 25 sub-LUTs corresponding to five kinds of canopy structure and five kinds of soil background. The average of the LCC inversion results of all sub-LUTs for each plant function type (PFT) was considered as the retrieval. The LUT algorithm was validated using the synthetic dataset, which gave an R2 value higher than 0.79 and an RMSE value lower than 10.5 μg cm−2. Then, the GLCC dataset was generated using the MERIS/OLCI multispectral data over 2003–2012 and 2018–2020 at a spatial resolution of 500 m and temporal resolution of one week. The GLCC dataset was validated using 161 field measurements, covering six PFTs. The validation results yielded an overall accuracy of R2 = 0.41 and RMSE = 8.94 μg cm−2. Finally, the GLCC dataset presented acceptable consistency with the existing MERIS LCC dataset developed by Croft et al. (2020). OLCI, as the successor to MERIS data, was used for the first time to co-produce LCC data from 2003–2012 to 2018–2020 in conjunction with MERIS data. This new GLCC dataset spanning nearly 20 years will provide a valuable opportunity for the monitoring of vegetation growth and terrestrial carbon cycle modeling. The GLCC dataset is available at https://doi.org/10.25452/figshare.plus.20439351 (Qian et al., 2022b).
BibTeX
@misc{qian2022global,
  author = {Xiaojin Qian and Liangyun Liu and Xidong Chen and Xiao Zhang and Siyuan Chen and Qi Sun},
  title = {{The global leaf chlorophyll content dataset over 2003–2012 and 2018–2020 derived from MERIS/OLCI satellite data (GLCC): algorithm and validation}},
  year = {2022},
  doi = {10.5194/essd-2022-277},
}

Loading related content...

Figure