Maximum likelihood and support vector machine for thematic maps classification in Bahr Al-Najaf, Iraq: Performance evaluation

Water scarcity is an emerging issue and the threat is mainly realized arid and semi-arid zones. Bahr AL-Najaf in Iraq is an example of the area with the shrinking level of the local water source, the lake is a major water resource for the population in the region. This paper analyzes Landsat and Sentinel - 2B satellite imagery and their role in surface water mapping in Bahr A-Najaf. It also evaluates the maximum likelihood classification (MLC) and Support Vector Machine (SVM) methods of classification to find out the most appropriate for surface water mapping. The project site is Bahr A- Najaf with the catchment area around it. Three images composed of one Sentinel-2B image (Landsat 5) for the year 2005 and two Sentinel-2B images (2015) and 2022 (2022) were used for the classification model tests. SVM and MLC methods were investigated over water classification surface and following. The accuracy assessment outcome was brought to light. SVM technique generally used more of its power to detect the higher accuracy and the reference data overlap in number than MLC did. Besides the wavy surface areas of Bahr Al-Najaf that were determined through classification maps, a study of their fluctuations was conducted. At the beginning, the total area is 63.59 square km (in 2005), then it was more than 10.77 square km (in 2015). At the ending, the area is even 32.77 square km (in 2022). The pragmatic application of SVM method captures the suitability of this method for the surface water classification; the accuracy and efficiency of it in the detailed mapping and control of surface water in Bahr AL-Najaf is undoubtedly more than satisfactory.