Sharma Megha, Goel Shailendra, Elias Ani A
Department of Botany, University of Delhi, Delhi, India.
ICFRE-Institute of Forest Genetics and Tree Breeding, Coimbatore, Tamil Nadu, India.
Sci Rep. 2025 Jan 2;15(1):44. doi: 10.1038/s41598-024-83551-9.
Evaluating high-throughput soil profile information is essential in safflower precision agriculture, as it facilitates efficient resource management and design of an experiment that promotes sustainable production. We collected soil from representative target environments (TE) of safflower cultivation and evaluated 14 soil physio-chemical features for constructing fine-resolution maps. The robustness, versatility, and predictive ability of two statistical learning models in correctly classifying the soil profile to clusters were tested. Calcium, sand, soil organic carbon, phosphorous, potassium, and sodium were found to be most influential in classifying the representative TE. Random Forest model was found to be the best performing with average prediction accuracy above 85% in all test settings which reached 100% in some. The optimal training population size for prediction was found to be 70-80%. The spatial distribution of sodium in Delhi was found to be aligned with the low yield of safflower emphasizing the importance of fine-resolution soil mapping to design a field experiment and optimize the nutrient supply. Fine-resolution mapping not only enhance soil management strategies but also support government initiatives such as soil health cards, delineation of cultivable land, and risk assessments in crop-growing areas.
评估高通量土壤剖面信息在红花精准农业中至关重要，因为它有助于高效的资源管理和促进可持续生产的实验设计。我们从红花种植的代表性目标环境（TE）中采集土壤，并评估了14种土壤理化特征以构建高分辨率地图。测试了两种统计学习模型在将土壤剖面正确分类到聚类中的稳健性、通用性和预测能力。发现钙、砂、土壤有机碳、磷、钾和钠在对代表性TE进行分类时最具影响力。发现随机森林模型表现最佳，在所有测试设置中平均预测准确率高于85%，在某些设置中达到100%。发现预测的最佳训练群体规模为70-80%。发现德里钠的空间分布与红花低产情况一致，强调了高分辨率土壤制图对设计田间实验和优化养分供应的重要性。高分辨率制图不仅能加强土壤管理策略，还能支持政府举措，如土壤健康卡、可耕地划定以及作物种植区的风险评估。
