Developing object-based image procedures for classifying and characterising different protected agriculture structures using LiDAR and orthophoto

Abstract

Controlled environment conditions inside protected agriculture (PA) structures can lead to the development of sustainable agriculture. In developed countries, the rapid growth of technology of sustainable, environmentally friendly agriculture via greenhouses or net-houses is due to the significant changes in climate and increasing demand for quality products such as vegetables, fruits, herbs, etc. Therefore, there is a need to map and classify different types of PAs to help understand the pattern of crop production. Using remote sensing, the mapping of PAs has gained significant consideration in recent decades. The main goal of this study is to develop a cost-effective, novel approach to create object-based image procedures for classifying and characterising different structures of PAs. To fulfil this goal, the project integrates high-resolution orthophoto and LiDAR data. Eleven distinctive major PA classes were identified, differing in size, height, construction, shape, materials and orientation. The research was conducted over a cluster of PAs, in the Arava Valley, Wadi Araba, Israel, and demonstrated an overall accuracy and Kappa index of agreement (KIA) 92% and 0.91, respectively. Remote information and discrimination of different types of structures within a PA cluster can provide important data to decision-makers, managers, environmental protection officers and others. Authorities might infer data about the number of farms, what is being cultivating and when, or, if the PA is abandoned. Such information can also be used for quantifying damage, for predicting the dispersion of virus and help strategic planning.