Appraising the Use of Remote and Proximal Platforms for Wheat Biomass Estimation from Multispectral Imagery

Remote and proximal sensing techniques have been demonstrated as useful tools for crop monitoring . The improvements in the platforms and sensors on board utilised for acquiring the imagery make it necessary to analyse how the resulting spatial and temporal resolutions influence crop biomass estimations. Thus, the main objective of the study was to evaluate the effectiveness of multispectral remote and proximal sensing imagery acquired from different platforms for estimating durum wheat biomass. This study conducted a comparison of multiple grain estimation methodologies across various image acquisition platforms, including unmanned aerial vehicle (UAV), and Sentinel-2 and Landsat8/9 satellites. The research methodology involved the collection and analysis of multispectral images during key stages of wheat growth. Wheat biomass was calculated using several empirical equations obtained from literature based on remote sensing indices calculation, such as Normalized Difference Vegetation Index, Normalized Difference Red-Edge Index and Soil Adjusted Vegetation Index. Such estimates were compared with the final crop yield measured on the plot. The findings revealed distinctive strengths and limitations between wheat estimation models and the most accurate phenological phase for performing the estimation. Differences among platforms were also observed. Specifically, UAV imagery provided a very high spatial resolution data that favoured precise, localized timing, at the expense of a reduction in its temporal resolution. On the contrary, Sentinel-2 and Landsat 8/9 offered broader coverage with moderate resolution, but with a higher revisit period. This comparative analysis highlights the importance of leveraging a combination of remote and proximal sensing technologies for an accurate and comprehensive estimation of durum wheat biomass. It highlights the need for tailored approaches based on spatial and temporal requirements, facilitating decision-making in agricultural management practices. Future research directions could explore further algorithm integration and refinement strategies to improve the accuracy of wheat crop biomass estimation and variety cultivar/species discrimination.