Utilizing Remote Sensing Data to Ascertain Weed Infestation Levels in Maize Fields

This study presents the evaluation of tools for weed analysis and management to support agroecological practices in organic farming, emphasizing agriculture digitalization and remote sensing. The main aim was to provide techniques for monitoring and prediction of weed spread using multispectral satellite and drone data, without the use of chemical inputs. Key findings indicate that VV and VH channels of Sentinel-1 and B2, B3, B4 and B8 channels of Sentinel-2 are not different regarding tillage, herbicide use, or sowing den-sity. However, RE and NIR channels of drone detected significant variations and proved effectiveness for weediness monitoring. The NIR channel is sensitive to agrotechnical factors such as cultivation type, making it valuable for field monitoring. Correlation and regression analyses revealed that B2, B3, B8 channels of Sentinel-2 and RE and NIR drone channels are the most reliable for predicting weed levels. Conversely, Sentinel-1 showed limited predictive utility. Random effect models confirmed that Sentinel-2 and drone channels can accurately account for site characteristics and timing of weed proliferation. Taken together these tools provide effective organic weed monitoring systems, enabling rapid identification of problem areas and adjustments in agronomic practices.