APP-TRI: foliar sampling in wheat to “read” the plot and refine the harvest prediction

Source: APP-TRI

The APP-TRI Operational Group is making progress on one of the project's key pillars: translating real-time agronomic data from the field into a digital application capable of estimating wheat productivity and anticipating potential crop deviations . The tool relies on integrating information from combine harvester yield maps and remote sensors (Sentinel-2) , with the aim of facilitating earlier management decisions —such as fertilization, disease control, or strategy adjustments—and improving efficiency and sustainability on cereal farms.

Within this framework, the Polytechnic University of Valencia (UPV) has developed, for dissemination, a document with images and descriptions of the sampling of plant material carried out in wheat plots in Burgos and Soria during the 2024-2025 season. The photographs capture different moments of the monitoring and have a common element: the sampling method, designed to be repeatable, objective and statistically sound.

A field sampling designed to minimize bias

The procedure begins when the evaluator enters the plot equipped with a rigid 40×40 cm frame and cutting tools. To avoid the "edge effect," the technician moves away from the field edges and selects points completely randomly. In practice, the technician walks across the plot in a zigzag pattern and, after a predetermined number of steps and without observing the soil, places the frame in the plot. The objective is clear: to ensure that the sample is representative and that the selection criteria are not influenced by the appearance of the crop.

Once the frame is in place—preferably diagonally across the rows to better capture spatial variability—a strict inclusion criterion is applied: only plants whose stems physically emerge within the perimeter are considered valid. Those plants that, even if they slant their leaves or ears inwards, root outside the delimited area are manually removed. The biomass is then cut as close to the ground as possible, or only the ears are collected, depending on the objective of the analysis. The material from both points is stored in properly labeled bags, and when combined, a final sample equivalent to 0.32 m² is obtained. Based on this, the density or yield results can be extrapolated to a hectare using expansion factors, assuming that the randomness of the sampling points compensates for the natural variability of the terrain.

According to the material provided by the UPV, three samplings were carried out in each province: in Burgos (April 25, May 28 and July 8, 2025) and in Soria (April 25, May 27 and July 9, 2025), capturing different stages of the crop throughout the cycle.

From the field to the laboratory: biomass, yield and nutrition

After collection, the samples were transferred to the laboratory for processing and characterization. An essential preliminary step is controlled dehydration to constant weight; the project's work plan includes oven drying at 65°C for 48–72 hours, ensuring that subsequent measurements are made on dry matter.

Based on this, biometric variables and yield components are quantified: number of stems and ears per unit area, average canopy height, total and ear dry matter weight, and thousand-grain weight as an indicator of grain filling and quality. In parallel, elemental composition is determined, differentiating leaf tissue and grain subsamples to measure carbon and nitrogen and derive physiological indicators, such as nitrogen use efficiency (NUE). This analytical block is complemented, within the framework of the project, by determinations of macro- and micronutrients and elemental analysis methods as outlined in the technical report.

Remote sensing: converting sampling points into decision maps

The ultimate goal is to link handheld measurements with the crop's spectral response: cross-referencing point data on biomass and nutritional status with satellite-derived vegetation indices (such as NDVI or GNDVI) to identify crop responses to treatments or agricultural practices. This correlation helps validate predictive models and transform discrete data into continuous maps of vigor and yield potential, a key component for the future APP-TRI app to be useful, interpretable, and applicable to individual fields.

The project also anticipates that monitoring the "showcase" farms will consolidate production data, soil analyses, foliar analyses, and climatological data into campaign reports. Foliar sampling is not a laboratory "extra": it is a direct reflection of what is happening in the field, with strategic value for anticipating decisions and strengthening the robustness of the models that will allow for early harvest predictions.