Retrieval of Harmonized LAI Product of Agricultural Crops from Landsat OLI and Sentinel-2 MSI Time Series

0567215 - ÚVGZ 2023 RIV CH eng J - Článek v odborném periodiku
Tomíček, J. - Mišurec, J. - Lukeš, Petr - Potůčková, M.
Retrieval of Harmonized LAI Product of Agricultural Crops from Landsat OLI and Sentinel-2 MSI Time Series.
Agriculture-Basel. Roč. 12, č. 12 (2022), č. článku 2080. E-ISSN 2077-0472
Grant CEP: GA TA ČR TH02030248
Institucionální podpora: RVO:86652079
Klíčová slova: Sentinel-2 * Landsat * leaf area index * harmonization * vegetation index * prosail * radiative transfer * artificial neural network * time series
Obor OECD: Agriculture
Impakt faktor: 3.6, rok: 2022
Způsob publikování: Open access
https://www.mdpi.com/2077-0472/12/12/2080

In this study, an approach for the harmonized calculation of the Leaf Area Indices (LAIs) for agronomic crops from Sentinel-2 MSI and Landsat OLI multispectral satellite data is proposed in order to obtain a dense seasonal trajectory. It was developed and tested on dominant crops grown in the Czech Republic, including winter wheat, spring barley, winter rapeseed, alfalfa, sugar beet, and corn. The two-step procedure harmonizing Sentinel-2 MSI and Landsat OLI spectral data began with deriving NDVI, MSAVI, and NDWI_1610 vegetation indices (VIs) as proxy indicators of green biomass and foliage water content, the parameters contributing most to a stand's spectral response. Second, a simple linear transformation was applied to the resulting VI values. The regression model itself was built on an artificial neural network, then trained on PROSAIL simulations data. The LAI estimates were validated using an extensive dataset of in situ measurements collected during 2017 and 2018 in the lowlands of the Central Bohemia Region. Very strong agreement was observed between LAI estimates from both Sentinel-2 MSI and Landsat OLI data and independent ground-based measurements (r between 0.7 and 0.98). Very good results were also achieved in the mutual comparison of Sentinel-2 and Landsat-based LAI datasets (rRMSE < 20%, r between 0.75 and 0.99). Using data from all currently available Sentinel-2 (A/B) and Landsat (8/9) satellites, a dense harmonized LAI time series can be created with high potential for use in precision agriculture.
Trvalý link: https://hdl.handle.net/11104/0338497