Geodaysit 2023

Pixel Mixture Issue in Mapping Vineyard Phenology. A Possible Solution Based on Sentinel-2 Imagery and Local Least Squares
06-12, 15:15–15:30 (Europe/London), Sala Biblioteca @ PoliBa

Precision viticulture aims to enhance quality standards of wine production by improving vineyard management. In this framework, satellite optical remote sensing has already proved to be effective for mapping vegetation behavior in space and time. These maps, properly processed, are useful to optimize agronomic practices improving wine production/quality and mitigating environmental impacts. Nevertheless, vineyards represent a challenge in this context because grapevine canopies are discontinuous, and the observed reflectance signal is affected by background. In fact, satellite imagery ordinarily provides spectral measures with medium-low geometric resolution (≥ 100 m2). Therefore, spectral mixture between grapevine canopies, grass and soils is expected within a satellite-derived reflectance pixel and not considering this problem can deeply affect deductions based on this data. In this work, Sentinel-2 (S2) NDVI maps (10 m resolution) were computed and compared to the ones obtained from DJI P4 multispectral UAV over a vineyard sizing 1.5 ha and located in Piemonte region (NW Italy). The proportion of row and inter-row (α(x,y) and 1-α(x,y)) within S2 pixel was computed and mapped classifying DJI photogrammetry point cloud. Involving α(x,y) and S2 NDVI values, reversing spectral unmixing system was defined solving for two average endmembers NDVI values (row and inter-row) using a moving window (21x21 pixels) least squares approach. Results were compared at S2 pixel-level to the average ones computed from DJI, showing a MAE of 0.15 and 0.10 of row and inter-row NDVI respectively.

Dott. Federica Ghilardi

PhD Student – SUSTNET
Department of Agricultural, Forest and Food Sciences - DISAFA - University of Torino (Italy)
GEO4Agri Lab - Laboratorio di Geomatica e Telerilevamento Agro-Forestale del DISAFA
Link: https://orcid.org/0000-0001-6447-9442

This speaker also appears in:

Dott. Alessandro Farbo

PhD Student - SUSTNET (Unito)

Department of Agricultural, Forest and Food Sciences - DISAFA - University of Torino (Italy)

GEO4Agri Lab - Laboratorio di Geomatica e Telerilevamento Agro-Forestale del DISAFA

This speaker also appears in:

Dr. Filippo Sarvia graduated with full marks from the University of Torino with a Master's degree in Agricultural Science. He won the annual award for best thesis in optical remote sensing (2019). Immediately after graduation, he won a scholarship and successfully competed for a Ph.D. position with DISAFA. Presently, his research objectives concern remote sensing technology for agroforestry. In particular, he is dealing with climate change-related topics, such as evaluation of the reaction of natural and agricultural systems to ongoing changes (drought, floods and hail); EU CAP controls by multi-temporal satellite imagery; and damage estimates to crops by extreme weather events (supporting insurance policies).

This speaker also appears in: