Monitoring and observing the movement of sand dunes and their impact on urban centers, engineering facilities, and agricultural lands in Al-Ahsa governorate using geo- AI and GIS for environmental sustainability

The study emphasizes the significance of geographic information systems (GIS) and geospatial artificial intelligence (GeoAI) in monitoring and predicting sand movement within Al-Ahsa Governorate. This is accomplished through the analysis of satellite imagery, field measurements, and artificial intelligence models to identify sand sources, determine annual movement rates, and assess their impact on urban areas, agricultural lands, and transportation infrastructure. By extrapolating the data, the Redness Rating results indicated that approximately 40% of the sand sources are continental, while 60% are marine, providing valuable insights for decision-makers to implement appropriate sand stabilization technologies. Additionally, the rates of sand encroachment are influenced by various geographical factors. The vegetation change detection index (NDVI) analysis from 2000 to 2020 revealed an increase in vegetation density from 0.12 to 0.52 in certain regions, attributed to afforestation, windbreaks, and agricultural activities. The terrain is characterized by flat, gently sloping plains ranging from 57 to 218 meters above sea level, with some isolated hills. Sand movement is predominantly driven by winds from the north, northeast, and northwest, resulting in higher sand concentrations in the northern parts of the study area. Regarding the performance of artificial intelligence models, the Random Forest (RF) model demonstrated superior accuracy compared to Support Vector Machine (SVM) and Artificial Neural Network (ANN) models when integrated with GIS indicators to monitor sand dune movement and predict their severity. The RF model achieved an accuracy of 0.937, with a strong Area Under the Receiver Operating Characteristic curve (AUCROC) of 0.982 and an Area Under the Precision-Recall curve (AUCPR) of 0.963. The study produced a predictive sand encroachment map indicating an average annual movement rate of 13 meters, with hazard zones classified as high, medium, or low for urban areas, agricultural lands, and roads. It recommends the application of deep learning models within GeoAI and GIS to enhance the accuracy of sand movement predictions. Furthermore, the study highlights the importance of utilizing predictive models and maps in selecting sand stabilization techniques such as afforestation, tree belts, and engineering methods. It also underscores the necessity of integrating GeoAI models and GIS indicators into environmental impact assessments to promote sustainable development planning.