A Micro-Scale Approach for Cropland Suitability Assessment of Permanent Crops Using Machine Learning and a Low-Cost UAV

This study presents a micro-scale approach for the cropland suitability assessment of permanent crops based on low-cost unmanned aerial vehicle (UAV) equipped with commercially available RGB sensor. The area was divided into two subsets, subsets A and B containing tangerine plantations planted during years 2000 2008, respectively. fieldwork performed 27 September 2021 by using Mavic 2 Pro UAV commercial in two-step classification process, utilizing: (1) supervised machine learning algorithms creating vegetation mask; (2) unsupervised according to Food Agriculture Organization United Nations (FAO) land standard. overall accuracy kappa coefficients were used assessment. most accurate combination input data parameters ANN all nine rasters, managing utilize complimentary information regarding spectral topographic properties. resulting levels indicated positive both 63.1% suitable subset 59.0% B. Despite that, efficiency agricultural production can be improved crop soil properties currently non-suitable class (N1), providing recommendations farmers further agronomic inspection. Alongside UAV, open-source GIS software globally accepted FAO standard are expected improve availability its application plantation management.