Systematic Review of Machine Learning Applications in Water Resources Management in Pakistan

Abstract

Almost all data-driven fields have been revolutionized by machine learning. Water resources management is a rich domain of data, and Machine learning can help improve water resource management by leveraging such a large amount of data. A plethora of studies worldwide have established significant potential in the application of machine learning in water resources management and its subfields. However, a relatively limited number of studies have been conducted in Pakistan, and the opportunity of machine learning from a Pakistani perspective has not been explored. This study aims to present a systematic review of machine learning application studies conducted in Pakistan in the area of water resources management. Over the past five years, nearly two dozen studies have been reviewed that have been done in Pakistan. Studies reviewed focus on groundwater quality and recharge monitoring and management, stream flow forecasting, climate change impacts on stream flows, surface water quality modeling, and drought prediction and management. Different machine learning algorithms were applied to get results for the above-mentioned tasks. For example, Random Forest, Decision Trees, Support Vector Machine (SVM), ANNs, and CNNs. The use of RNN-LSTM and IoT are getting popular day by day. Different researchers explored different machine learning algorithms to perform the same task, and compared the performances of these algorithms. The current study also seeks to suggest a new direction for machine learning-based water resource management in Pakistan.