Bibliometric analysis of machine learning applications in remote sensing of soil salinization

In recent years, with the development of machine learning algorithms, national and international experts and researchers have devoted themselves to the study of remote sensing of soil salinization using machine learning models and have achieved fruitful results. Using the bibliometric visualization software CiteSpace, this study reviewed the research themes of machine learning-based remote sensing modeling of soil salinization in the last decade. The main research themes from recent years were summarized, and research progress around machine learning algorithms, modeling feature variables, and model evaluation were discussed. Additionally, the study examined the limitations and development trends around current research themes. The main conclusions were: 1) the main research themes included machine learning algorithms and their accuracy, modeling feature variable selection, the impact of remote sensing data source selection on models, selection of study areas for soil salinization, and the application of machine learning-based digital mapping of soil salinization. 2) The current research themes were the application of covariates as feature variables in model construction, the combination of measured spectral data and multi-source remote sensing spectral data, and the selection of the most effective machine learning algorithms. 3) Based on 2018, research progress could be divided into the initial stage and the high-speed development stage. Soil salinization remote sensing monitoring based on cloud platforms and machine learning will become a direction for future research development.