The Application of Smart and Precision Agriculture (SPA) for Measuring Leaf Nitrogen Content of Oil Palm in Peat Soil Areas

One of the laborious tasks in operating and managing oil palm lands is measuring the nutrient content of leaves in a timely and periodic manner. It takes a manual process and takes a long time to get leaf samples that are representative of the observed oil palm area, especially in peat soil area. In addition, laboratory analysis of leaf samples is expensive and introduces creates significant time delays for queuing. This paper discusses methods for measuring leaf nitrogen levels in a timely manner based on imagery obtained by C-SAR Sentinel-1A using smart and precision agricultural approaches and applications. The intelligent computational model uses Random Forest Regression to calculate leaf nitrogen content based on spatial image processing and analysis. The calculated leaf nitrogen content is then used to examine and identify the variability of nitrogen content in a given oil palm area. Identification of nutrient variability is one of the approaches and goals of precision agriculture. Our computation model was developed using 553 leaf sampling units (LSU) and validated using 10-Fold validation method, resulting in correctness, MAPE, RMSE and R2 of 95.69%, 4.31%, 0.1119 and 93.28% respectively. The validated model was tested using a set of new data, which is not in the data samples for model development and validation, was resulting in correctness, MAPE, RMSE and R2 of 89.91%, 10.11%, 0.284, and 53.89% respectively. The visual maps can be produced and displayed to visualize variability of nitrogen contents in a peat soil area. This improves visibility and transparency in monitoring nutrient in timely and periodic manner. The future use of the proposed method is to support the application of smart and precision agriculture practices to recommend fertilizers types and dozes for oil palm in peat soil areas based on the variability of leaf nutrient contents.