Forage Biomass Estimation Using Sentinel-2 Imagery at High Latitudes

• Published in: Remote Sensing
• Main Authors: PENG Junxiang, ZEINER Niklas, PARSONS David, FÉRET Jean-Baptiste, SÖDERSTRÖM Mats, MOREL Julien
• Language: English
• Published: MDPI 2023
• Subjects: Nash; Sutcliffe; Northern Sweden
• Online Access: https://publications.jrc.ec.europa.eu/repository/handle/JRC135760; https://www.mdpi.com/2072-4292/15/9/2350; https://doi.org/10.3390/rs15092350

Forages are essential crops in high-latitude regions and the primary feed source for ruminant-based dairy industries. Maximizing farm economic and ecological performance, as well as meat and dairy sector performance, requires timely and relevant field-specific data, such as available biomass. Sentinel-2 satellites provide open-access spectral data for frequent vegetation monitoring. This spectral data was used to estimate the in-season dry matter yield (DMY) of harvested forage fields in northern Sweden. Univariate and multivariate regression models, including partial least squares, support vector machines, and random forests, were tested for their ability to accurately and robustly estimate DMY using reflectance values and vegetation indices derived from Sentinel-2 spectral bands. Random forest regression (RFR) produced the most stable and robust results, with Nash-Sutcliffe model efficiency (NSE) values of 0.92, 0.55, and 0.86 for calibration, validation, and evaluation, respectively. These results suggest that RFR can be used to accurately estimate DMY of forage fields using Sentinel-2 data. However, larger and more comprehensive datasets are needed to validate these results and to assess the performance of RFR in other regions and under different conditions. JRC.D.5 - Food Security