Periodic Reporting for period 1 - MOVES (MOnitoring VEgetation status and functioning at high spatio-temporal resolution from Sentinel-2)
Periodo di rendicontazione: 2019-10-21 al 2021-10-20
The status and functioning of vegetation play key roles in many biophysical processes, such as photosynthesis and transpiration, and therefore significantly influence the global climate system and carbon cycle. Accurate estimation of canopy biophysical parameters is the prerequisite for a wide range of agricultural, ecological, hydrological and meteorological applications. LAI, FAPAR and FCOVER are three widely used biophysical parameters. LAI is defined as one-half the total area of green leaves per unit area of horizontal ground, and it controls the exchanges of energy, water, and greenhouse gases between the land surface and the atmosphere. FAPAR is dened as the fraction of radiation absorbed by the canopy in the 400-700 nm spectral domain and is a main input in models of light-use efficiency to estimate Gross Primary Productivity (GPP). The FCOVER, which is defined as the fraction of the background covered by green vegetation as seen from the nadir, is also a pertinent variable that can be used in models of the surface-energy balance to separate the contribution of the soil from that of the canopy.
High spatiotemporal resolution LAI/FAPAR/FCOVER products are urgently needed to characterize terrestrial ecosystem processes, supporting crop management activities at the parcel level and for better understanding land greening and browning phenomena at local scales. The overall objective of this project is to develop a robust, efficient and highly accurate LAI/FAPAR/FCOVER retrieval algorithm for Sentinel-2 data and therefore achieve the operational monitoring of vegetation structure and functioning at a high spatiotemporal frequency.
Due to the adverse impact of Covid-19, it is very hard to implement the project fully following the WP. For example, the time spending on field measurement, secondment and conference was significantly reduced because of the strict confinement. However, our project still achieved fruitful results. Main project results include:
(1) Proposed an integrated method for Sentinel-2 topographic and angular normalization, which could estimate the intrinsic vegetation reflectances and improve retrieval accuracy of biophysical parameters.
(2) Validated decametric (Sentinel-2) and kilometric LAI (MODIS, CGLS, SAF, GLASS and C3S) product in Maize Crops, which informed the retrieval algorithm improvement.
(3) Proposed a multiband constraint harmonization method for combined use of Sentinel-2 and Landsat 8, which contributes to generate spatiotemporally continuous time series of decametric data from the MSI-OLI virtual constellation.
(4) Developed an algorithm to generate daily 16-m biophysical parameter maps by the combination of LAINet observation system and Gaofen-1/6 satellite images, which is crucial for agricultural monitoring.
(5) We also compared the performance of structure and physiology vegetation indices for photosynthetic phenology monitoring, and extend the application of CCI (a reliable proxy for tracking GPP dynamics) to sentinel-2 by replacing narrowband green with broadband green.
The results of the above work have been and will further be made available to targeted peers:
(1) Parts of results have been published in top-ranking journal on EO (6 SCI papers, with 1 on Geophysical Research Letters, 1 on IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 1 on IEEE Geoscience and Remote Sensing Letters, and 3 on Remote Sensing), and others are in review (1 on Photogrammetric Engineering and Remote Sensing and 1 on Journal of Remote Sensing) or preparation.
(2) A project website was successfully established: https://www.creaf.cat/monitoring-vegetation-status-and-functioning-high-spatio-temporal-resolution-sentinel-2
(3) Our results will inform the decision-making for the stakeholders in agriculture, forestry, and insurance.
High spatiotemporal resolution LAI/FAPAR/FCOVER products are urgently needed to characterize terrestrial ecosystem processes, supporting crop management activities at the parcel level and for better understanding land greening and browning phenomena at local scales. The overall objective of this project is to develop a robust, efficient and highly accurate LAI/FAPAR/FCOVER retrieval algorithm for Sentinel-2 data and therefore achieve the operational monitoring of vegetation structure and functioning at a high spatiotemporal frequency.
Due to the adverse impact of Covid-19, it is very hard to implement the project fully following the WP. For example, the time spending on field measurement, secondment and conference was significantly reduced because of the strict confinement. However, our project still achieved fruitful results. Main project results include:
(1) Proposed an integrated method for Sentinel-2 topographic and angular normalization, which could estimate the intrinsic vegetation reflectances and improve retrieval accuracy of biophysical parameters.
(2) Validated decametric (Sentinel-2) and kilometric LAI (MODIS, CGLS, SAF, GLASS and C3S) product in Maize Crops, which informed the retrieval algorithm improvement.
(3) Proposed a multiband constraint harmonization method for combined use of Sentinel-2 and Landsat 8, which contributes to generate spatiotemporally continuous time series of decametric data from the MSI-OLI virtual constellation.
(4) Developed an algorithm to generate daily 16-m biophysical parameter maps by the combination of LAINet observation system and Gaofen-1/6 satellite images, which is crucial for agricultural monitoring.
(5) We also compared the performance of structure and physiology vegetation indices for photosynthetic phenology monitoring, and extend the application of CCI (a reliable proxy for tracking GPP dynamics) to sentinel-2 by replacing narrowband green with broadband green.
The results of the above work have been and will further be made available to targeted peers:
(1) Parts of results have been published in top-ranking journal on EO (6 SCI papers, with 1 on Geophysical Research Letters, 1 on IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 1 on IEEE Geoscience and Remote Sensing Letters, and 3 on Remote Sensing), and others are in review (1 on Photogrammetric Engineering and Remote Sensing and 1 on Journal of Remote Sensing) or preparation.
(2) A project website was successfully established: https://www.creaf.cat/monitoring-vegetation-status-and-functioning-high-spatio-temporal-resolution-sentinel-2
(3) Our results will inform the decision-making for the stakeholders in agriculture, forestry, and insurance.
Rugged terrain is a major source of uncertainty for biophysical retrieval over mountainous areas. Traditionally methods often rely on radiative transfer model to simulate vegetation reflectances over mountainous areas. However, the complex model structure and the parametrization hinder its operational use. During the reporting period, ER proposed to convert the vegetation reflectances over rugged terrain (ρo) to their counterpart over flat terrain (ρc) through a multiplied operator (P), i.e. ρc = ρo × P. P was derived from the simplification of an our previously proposed Path Length Correction (PLC) model. This method is characterized by mathematical simplicity and physical soundness, and provides an efficient tool to correct the terrain-induced canopy BRDF distortion and will facilitate the exploitation of multi-angular information for biophysical parameter retrieval over mountainous regions.
Representativeness is the prerequisite for improving the retrieval accuracy of biophysical parameter from remote sensing data. We collected field measurements from multi-instruments, including LAI-2000, digital hemispherical photography, and wireless sensor network, and constructed a training dataset with high representativeness. A machine learning (specifically, random forest) model was trained to link the collected field measuremts with corresponding reflectances. These technologies support us generating LAI/FAPAR/FCOVER maps of decametric resolution with daily revisiting frequency. In addition, we proposed a new threshold-based phenology extraction method which is directly applied to daily nonsmoothed time series without any additional preprocessing steps. This method is robust and computationally efficient for high spatiotemporal phenology monitoring.
With the help from our Chinese colleagues, we refined our wireless sensor network-based unattended field measurement system to support temporally continuous automatic measurement. We also developed a ground measurement method based on smartphone camera, which significantly improve the flexibility of the field work. With the collected field dataset from our newly developed instruments, we validated several decametric (Sentinel-2) and kilometric LAI (MODIS, CGLS, SAF, GLASS and C3S) products in Maize Crops.
Representativeness is the prerequisite for improving the retrieval accuracy of biophysical parameter from remote sensing data. We collected field measurements from multi-instruments, including LAI-2000, digital hemispherical photography, and wireless sensor network, and constructed a training dataset with high representativeness. A machine learning (specifically, random forest) model was trained to link the collected field measuremts with corresponding reflectances. These technologies support us generating LAI/FAPAR/FCOVER maps of decametric resolution with daily revisiting frequency. In addition, we proposed a new threshold-based phenology extraction method which is directly applied to daily nonsmoothed time series without any additional preprocessing steps. This method is robust and computationally efficient for high spatiotemporal phenology monitoring.
With the help from our Chinese colleagues, we refined our wireless sensor network-based unattended field measurement system to support temporally continuous automatic measurement. We also developed a ground measurement method based on smartphone camera, which significantly improve the flexibility of the field work. With the collected field dataset from our newly developed instruments, we validated several decametric (Sentinel-2) and kilometric LAI (MODIS, CGLS, SAF, GLASS and C3S) products in Maize Crops.
The grant provided the ER the opportunity to join the world-class research group on Earth Observation and Global Change Ecology. Through this project, the ER further improved his abilities in vegetation radiative transfer modelling, the retrieval of biophysical parameters, and product validation. In addition, the supervisor (Profs. Peñuelas and Verger) helped the ER open new research lines (photosynthetic phenology, high spatio-temporal resolution biophysical parameter retrieval from EO data). The skills acquired by the researcher during this fellowship exerted a direct impact on his career development, transforming him into a completely independent research scientist. The ER has successfully obtained a tenured position in China after this fellowship.
The research results achieved during the grant period would be successively published on scientific journal, and social media (e.g. Twitter, Youtube). In future, the ER will devote himself to becoming a messenger to promote the exchange of science and technology between the European Union (EU) and China.
The research results achieved during the grant period would be successively published on scientific journal, and social media (e.g. Twitter, Youtube). In future, the ER will devote himself to becoming a messenger to promote the exchange of science and technology between the European Union (EU) and China.