Logo image
Machine learning applications for  wind resource mapping in Ajman,  UAE: towards sustainable energy  solutions
Dissertation   Open access

Machine learning applications for wind resource mapping in Ajman, UAE: towards sustainable energy solutions

Kaise Kunyil Muhammed Fasel
Doctor of Philosophy (PHD), University of Greater Manchester
23/05/2025

Abstract

Accurate site-specific Sustainable Wind Resource Assessment (SWRA) remains a critical contemporary sustainable wind power development issue. Especially in regions like the Emirate of Ajman and the United Arab Emirates (UAE), site-specific wind data collection faces high challenges and constraints. Mainly due to the excessively high costs of measuring wind speeds at wind turbine hub heights level, leading to dependence on publicly available NASA satellite data or any other freely available wind data that requires extensive error correction for reliable application in SWRA. This research develops a comprehensive methodology for site-specific SWRA in the Emirate of Ajman through five integrated objectives: developing machine learning (ML)-based error correction methodology for NASA satellite wind data, determining site-specific surface parameters, predicting future wind speed trends using ARIMA modelling, analysing wind potential variations, and creating GIS-based wind resource maps. A systematic mixed-methods approach was used, integrating multiple ML algorithms (Random Forest, Support Vector Machine, Gradient Boosting) for NASA wind speed data correction, determination of site-specific parameters (wind shear coefficients, roughness length, air density), statistical analysis of wind patterns, and GIS-based wind resource mapping. Ground-based measurements from strategically located onshore monitoring stations validated the methodology and established site-specific correction factors across Ajman's diverse terrain. Results showed clear spatial and temporal variations in wind resources, with annual wind speeds ranging from 3.33- 3.74 m/s at 50m to 4.75-5.2 m/s at 100m height. Spring emerged as the optimal season, with wind speeds reaching 5.69-6.16 m/s at 100m height. The Random Forest model achieved the highest accuracy (R² = 0.5772) in satellite data correction. Surface roughness length varied from 0.0002 (offshore) to 0.50 (urban areas), while air density ranged between 1.146-1.166 kg/m³. Offshore locations showed higher wind power density, reaching 126.12 W/m². This study establishes Ajman's first validated, GIS integrated SWRA methodology, contributing to practical and theoretical advances in SWRA. While supporting the feasibility of hybrid wind-solar systems and offshore installations, the findings align with the UAE's Net Zero 2050 strategy and establish a systematic approach that other regions can follow to improve satellite-derived wind speed data accuracy.
pdf
Machine Learning Applications for Wind Resource Mapping in Ajman, UAE5.72 MBDownloadView

Metrics

13 File views/ downloads
41 Record Views

Details

Logo image

Usage Policy