PRISMA and EnMAP comparison in the context of wheat nitrogen status assessment
(2023) 14th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS 2023) — Location: Athens, Greece (31.October.2023)
Files
WHISPERS2023Poster.pdf
Open Access - Adobe PDF
- 3.91 MB
Details
- Authors
- Abstract
- Over the past few decades, spaceborne hyperspectral imagery has evolved significantly, opening up new possibilities in the realm of agricultural research and monitoring. This technology has witnessed remarkable advancements in agricultural applications, such as crop type mapping and vegetation classification. In addition, hyperspectral data has proven invaluable in detecting biotic and abiotic stresses in agricultural systems. Furthermore, the retrieval of biophysical traits such as leaf area index, chlorophyll content and nitrogen concentration through hyperspectral data analysis has contributed to a deeper understanding of crop health and growth dynamics. Through the use of spaceborne hyperspectral imagery, agricultural stakeholders can make data-driven decisions, optimize agricultural practices, and ensure the sustainable production of food. The following experiment was carried out as part of the SPAGHYTI project, a collaboration led by CONSTELLR, a leading provider of space-based water, carbon, and temperature insights. CONSTELLR's mission is to advance understanding of the planet's ecosystems, enabling a more equitable and sustainable future governance of global resources. The SPAGHYTI project, conducted under the aegis of the SKYWIN competitiveness cluster, aims to develop field-level applications that deliver pertinent and actionable information to farmers. Specifically, the two applications are dedicated to the nitrogen status assessment and the monitoring abiotic and biotic stress levels of winter wheat in the Walloon region of Belgium. These applications leverage hyperspectral satellite imagery and aim to provide affordable access to high-quality information. The SPAGHYTI project encompasses a comprehensive value chain, covering sensor technology to end-user implementation. This paper reports a methodology built on the analysis of EnMAP and PRISMA data with the aim of assessing the added value of hyperspectral data for estimating the nitrogen content of wheat crops. An intensive field campaign was carried out from March to July 2023, i.e., during the nitrogen amendment period. Samples were taken from trial plots and several farmers' plots. These plots have been selected to match with the extent of the EnMAP and PRISMA images acquired. For each of these plots, dry matter and nitrogen content measurements were carried out in the laboratory. In addition, spectral data were acquired for each sampling unit using field portable spectroradiometer (ASD FieldSec 4). The first step of the study focuses on quantifying signal to noise ratio and its impacts on vegetated surfaces. Pure pixels of vegetation and invariants were selected in order to perform statistical analysis in both cases. Different wavelengths, wavelength combinations and spectral indices were then selected as the most relevant for estimating the nitrogen status of wheat by computing their correlation with the ground truth measurements of canopy nitrogen content (CNC). Different methods for estimating the nitrogen status of wheat were assessed according to their performances. Eventually, two different methods have been designed and implemented. The two methods rely on the estimation of dry matter (DM) and canopy nitrogen content (CNC) in order to produce a Nitrogen Nutrition Index (NNI). The first method relies on the use of hyperspectral vegetation indices (HVI) to estimate CNC and DM with simple linear regression. The second method relies on biophysical variables retrieved from physically-based model that have a direct correlation with the variables of interest. These performance results obtained from EnMAP and PRISMA are then compared to operational nitrogen status products derived from Sentinel-2 MSI data and the potential complementarity based on frequent Sentinel-2 data and precise hyperspectral retrieval is investigated.
- Affiliations
Citations
Troiani, M., Bouchat, J., Leclère, L., Curnel, Y., Vermeulen, P., Stevens, F., Scaut, B., Malice, D., Baeten, V., Chamberland, N., Planchon, V., & Defourny, P. (2023). PRISMA and EnMAP comparison in the context of wheat nitrogen status assessment. 14th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS 2023), Athens, Greece. https://hdl.handle.net/2078.5/213525