Title: Defining Agricultural Holdings in the Najd Region of Dhofar Governorate: An Integrated GeoAI, GIS, and Remote Sensing Approach
Short Biography:
Mariya Al Balushi is a Management Representative at Bayanat Geographical Consulting (BGC), her role combines operational leadership with a strong focus on business performance, collaboration, and sustainable growth.
With multidisciplinary experience, Mariya brings a versatile and commercially minded perspective to complex business environments as a unique blend of strategic thinking, commercial awareness, and multidisciplinary experience. Beyond her professional expertise, she is a Woman Candidate Master (WCM) in chess, an achievement that reflects her strategic thinking, discipline, patience, and ability to make informed decisions under pressure. She brings these same qualities to her professional practice, combining analytical insight with a forward-thinking approach to problem-solving and business coordination.
Abstract:
Accurate, up-to-date knowledge of agricultural land is central to food security, water governance, and equitable land administration, yet cadastral systems in many regions remain fragmented, manual, and slow to detect encroachment or change.
This paper presents an integrated GeoAI, GIS, and remote sensing framework developed to define and monitor agricultural holdings across the Najd region of Dhofar Governorate, Sultanate of Oman, commissioned by the Ministry of Housing and Urban Planning. Combining cadastral records, differential GPS field surveys, and multitemporal satellite imagery from 2009 and 2024, the project trained and deployed deep learning models, including UNET, Mask-RCNN, Vision Transformers, and the Segment Anything Model, to delineate farm boundaries and classify land use within an enterprise PostgreSQL/PostGIS, QGIS, and GeoServer environment. Across 13,362 km² and 1,894 surveyed points, the analysis identified 1,196 agricultural holdings and recorded a 77.7 percent increase in registered agricultural land, from 192.8 million to 342.7 million square meters between 2009 and 2024.
Delivered within three months despite years of prior delay, the resulting platform replaces fragmented manual records with an accurate, automated, and scalable geospatial system, offering a replicable model for modernizing agricultural land administration in arid and semi-arid regions.