Lizbeth Camacho Olivares


Session

11-12
11:45
15min
Analysis of fertilization in broccoli crops through precision agriculture techniques and edaphic parameters
Lizbeth Camacho Olivares

Introduction

It is estimated that by 2050, the global population will reach 9.73 billion, requiring food production to increase by more than 50% from 2012 levels (FAO, 2025). However, the growing global demand for food is driven not only by population growth but also by rising per capita income and economic development, which promote diets with higher caloric intake and greater consumption of animal protein (Tilman et al., 2011). In this context, the expansion of agricultural land is no longer considered a sustainable option because of its impacts on ecosystems and natural resources. Therefore, sustainable intensification seeks to increase agricultural production on existing croplands through technologies that optimize the use of key resources for productivity, particularly fertilizers (Foley et al., 2011).

In this context, fertilization is one of the agronomic practices with the greatest influence on crop yield because an adequate nutrient supply regulates essential physiological processes involved in plant growth and development, thereby improving agricultural productivity (Marschner, 2012; Tilman et al., 2011). However, in many agricultural production systems, fertilizers are still applied uniformly without considering crop requirements or the spatial variability of soil properties. As a result, some areas may receive insufficient fertilizer to meet crop nutrient demands. In contrast, others receive amounts that exceed the plants' uptake capacity, reducing nutrient use efficiency and increasing production costs (Havlin et al., 2014).

To optimize fertilization management, precision agriculture has incorporated remote sensing technologies that enable high-spatial-resolution monitoring of crop status. Among these, unmanned aerial vehicles (UAVs) equipped with multispectral sensors have proven to be an effective tool for estimating variables related to plant vigor, chlorophyll content, and nutritional status through vegetation indices derived from crop reflectance (Yang et al., 2025). Furthermore, the incorporation of machine learning techniques has improved nutrient management by enabling the integration and processing of large volumes of data and the solution of highly complex nonlinear problems. This enables automated, real-time diagnosis of crop nutritional status, enhancing precision and allowing producers to accurately predict plot yields and recommend optimal fertilization rates (Ennaji et al., 2023).

Although these technologies have shown promising performance for assessing fertilization status, most studies have focused on estimating specific nutrient concentrations, chlorophyll content, plant vigor, or crop yield. Consequently, there remains a need to develop methodologies that integrate edaphic information and spectral indices to jointly characterize nutritional status under real production conditions. In this context, the present study proposes developing a machine learning model that integrates variables derived from laboratory analyses and multispectral imagery acquired with a Parrot Sequoia camera to classify the fertilization status of broccoli crops into three conditions: deficiency, sufficiency, and over-fertilization.

Materials and Methods

The study is conducted in a 6-hectare pilot plot located in the municipality of Morelos, Zacatecas. Three 15 × 5 m polygons were delineated, upon which three fertilization treatments were established: deficiency (suspension of fertilizer application); sufficiency (conventional dose used by the farmer); and over-fertilization (application of double the conventional dose). The three experimental plots are expected to provide sufficient information to analyze at least 18 hectares of broccoli crops in subsequent growing seasons.

To characterize the soil nutrient status, georeferenced sampling and laboratory analyses were carried out to determine the concentrations of nitrogen (N), phosphorus (P), and potassium (K). Simultaneously, during the phenological stages of the crop, multispectral images were acquired using an unmanned aerial vehicle equipped with a Parrot Sequoia camera. From these images, vegetation indices related to crop nutritional status were calculated, including NDVI, SAVI, NDRE, and CIred-edge (Red-Edge Chlorophyll Index ​.

Subsequently, the edaphic variables and spectral indices were integrated into a database to train and evaluate different supervised machine learning algorithms. Finally, the performance of the models was assessed using classification metrics to determine their ability to discriminate among deficiency, sufficiency, and over-fertilization conditions in broccoli crops.

Expected Results

The model is expected to classify fertilization status in broccoli crops under deficiency, sufficiency, and over-fertilization conditions by integrating edaphic variables and spectral indices. Likewise, it is expected to identify the variables with the greatest discriminative power for each fertilization condition. Finally, thematic maps showing the spatial distribution of fertilization status across the experimental plots are expected to be generated, providing valuable information for decision-making on fertilization management.

Final Considerations

The proposed methodology integrates edaphic information, multispectral imagery, and machine learning techniques to classify the fertilization status of broccoli crops under real-world production conditions. This approach is expected to contribute to the development of decision-support tools in precision agriculture, with the potential to spatially identify zones of deficiency, sufficiency, and over-fertilization. The information generated could foster more efficient fertilizer management, optimize the use of available resources, and help reduce the environmental impacts associated with inappropriate application, thereby strengthening strategies for more sustainable agricultural intensification.

References
Ennaji, O., Vergütz, L., & El Allali, A. (2023). Machine learning in nutrient management: A review. Artificial Intelligence in Agriculture, 9, 1–11.
Foley, J. A., Ramankutty, N., Brauman, K. A., Cassidy, E. S., Gerber, J. S., Johnston, M., & Zaks, D. P. M. (2011). Solutions for a cultivated planet. Nature, 478(7369), 337–342.
Food and Agriculture Organization of the United Nations. (2025). The State of Food and Agriculture 2025: Addressing land degradation across landholding scales. FAO.
Havlin, J. L., Tisdale, S. L., Nelson, W. L., & Beaton, J. D. (2014). Soil fertility and fertilizers: An introduction to nutrient management (8th ed.). Pearson.
Marschner, P. (2012). Marschner’s Mineral Nutrition of Higher Plants (3rd ed.). Academic Press.
Tilman, D., Balzer, C., Hill, J., & Befort, B. L. (2011). Global food demand and the sustainable intensification of agriculture. Proceedings of the National Academy of Sciences of the United States of America, 108(50), 20260–20264.
Yang, X., Chen, J., Lu, X., Liu, H., Liu, Y., Bai, X., Qian, L., & Zhang, Z. (2025). Advances in UAV remote sensing for monitoring crop water and nutrient status: Modeling methods, influencing factors, and challenges. Plants, 14(16), 2544.

Remote Sensing, Photogrammetry, and Geodesy
Sala Geomática 7