Assessing the impacts of anthropogenic-induced land use/land cover changes in wetlands using remotely sensed information: A systematic state-of-the-art review and future directions

Abstract

Researchers have consistently strived to improve the discriminability of various land cover types, particularly between aquatic and vegetative areas, through diverse remote sensing techniques. This is crucial for conserving wetlands that have been degraded and undergone land use changes due to anthropogenic activities. This paper performs a meta-analysis and provides a systematic review of studies related to the land use classification process. After an extensive search, 74 papers were selected (PRISMA method) for detailed analysis. The study aims to introduce, investigate, and evaluate remote sensing methodologies for wetland land use classification and assess how these methods impact the detection of wetland change patterns. The findings reveal that 10 % of the studies utilized an object-based approach with optimization based on a trial-and-error method. Additionally, wetland researchers prefer combining Landsat data with supervised machine learning classification methods (82 %). This paper suggests conducting an in-depth examination of integrating dynamic training sample selection methods with object-based approaches, automating the optimization of segmentation parameters, and employing transfer learning techniques for classification. Moreover, the review highlights existing gaps and proposes future research avenues to advance research. For instance, improving accuracy is possible through explainable artificial intelligence and replacing the weak and commonly used Kappa with new evaluation metrics. Additionally, a new concept framework, “Aquatic Harmony/Aquatic Disruption,” and a wetland risk assessment map have been introduced, offering a comprehensive perspective on the impact of anthropogenic activities on wetlands. This review may open new horizons for wetland researchers by providing alternative approaches for the future.