Implementasi Precision Agriculture melalui Penyusunan Rekomendasi Pemupukan Tanaman Kelapa Sawit Berbasis Sistem Informasi Geografis (SIG) dan Teknologi Penginderaan Jauh

Abstract

This study evaluated methods for accurate formulation of fertilization recommendations for oil palm plantations in North Sumatra Province for implementing precision agriculture. The study used geographic information systems (GIS) and remote sensing technology to achieve this aim. Specifically, the study's objective is to develop estimates of the nutrient content of oil palm leaves using geospatial analysis and digital image processing of Landsat-8 data. These estimates were used to formulate fertilization recommendations for oil palms, and visualized in a web-GIS application. Methods used include: (1) Oil palm leaf nutrient classification through machine learning algorithms, such as support vector machines (SVM), random forests (RF), and classification and regression trees (CART), both executed in the Google Earth Engine (GEE) platform. (2) Spatial analysis using ordinary kriging (OK), universal kriging (UK), inverse distance weighted (IDW), and radial basis function (RBF) methods. (3) Nutrient balance assessment using the diagnosis and recommendation integrated system (DRIS). (4) Leaf nutrient content estimation was through vegetation indices, include: normalized difference vegetation index (NDVI), green normalized difference vegetation index (GNDVI), soil adjusted vegetation index (SAVI), modified soil adjusted vegetation index (MSAVI), optimized soil adjusted vegetation index (OSAVI), enhanced vegetation index (EVI), chlorophyll vegetation index (CVI), chlorophyll index green (CIG), normalized green red difference index (NGDRI), green leaf index (GLI), visible atmospherically resistant index (VARI), excess green (ExG), excess blue (ExB), excess red (ExR), and colour index vegetation extraction (CIVE). Furthermore, the data were compiled using PostgreSQLPostGIS; the web development was completed using Lavarel, and the web-GIS view was developed using Leaflet JS. The study provides a classification using the random forest (RF) and classification and regression tree (CART) algorithms with accuracy values over 90%. This geospatial analysis using OK, UK, IDW, and RBF interpolation methods showed similar results, with an accuracy above 90%. The DRIS index indicated that the average leaf nutritional requirements were in the following order: potassium (K), phosphorus (P), nitrogen (N), magnesium (Mg), and calcium (Ca), with corresponding values of -7.73, 0.03, 1.09, 1.94, and 4.41. The MSAVI and GNDVI indices demonstrated largest significant correlation between the vegetation indices and Mg nutrient content, with a value of r>0.8 for N content and between 0.6 and r>0.8 for P, K, Ca, and Mg nutrient content. The web-GIS provided a visualization of the database, and fertilizer recommendation dashboard. This system will assist in optimizing oil palm fertilizer recommendations in North Sumatra to increase yield and productivity.