**5. Conclusions**

In the present study, we demonstrate that field surveys in mangrove ecosystems are not always feasible, due to the high costs and inaccessibility of the area. Mangrove distribution mapping is a hot topic in the field of mangrove remote sensing [14]. Based on field observations, the mangrove forests in the present study have a uniform composition of the species *Avicenna marina* and the detectable di fferences are limited to canopy density, which consists of mangrove zonation patterns including forests of immature trees and of mature trees, and isolated trees. The use of VHR satellite imagery for sampling reference data in combination with freely available satellite data and machine learning is an e ffective and straightforward approach to further improve the details of land cover maps and to assess relevant forest parameters. Upscaling is a cost-e fficient tool for producing accurate large-scale land cover maps in inaccessible ecosystems. The findings of the present study support the sustainable managemen<sup>t</sup> of mangrove ecosystems and can be used to assess the e fficiency of ecosystem services. Although the upscaling approach produced low user accuracies for the shallow water and tidal zone classes, overall accuracies were generally high.

With the proposed method, it is possible to distinguish between the two most relevant classes for management, i.e., canopy mangrove canopy and mudflat. Our findings confirm that advances in remote sensing data and techniques are favorable for developing novel methods to map mangrove ecosystems in greater detail. We conclude that the selection of appropriate images remains an important factor and that Sentinel-2 images have grea<sup>t</sup> potential for identifying di fferent land cover types, thanks to their high spatial, temporal and spectral resolution. Continuity of the presented approach is guaranteed since Sentinel-2 data will be continuously acquired.

**Author Contributions:** N.B.T. is responsible for the study. N.B.T. developed the methodology, programming, statistical analysis and was the main writer of the manuscript. L.T.W. supported the development of the approach and methodology and writing the manuscript. C.G. supported the development of the approach and methodology. A.R.S., S.F. and S.P. conceived and provided useful suggestions to the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** This study was carried out in the framework of the doctoral thesis and was supported by the Department of Natural Resources, Isfahan University of Technology, Isfahan, Iran. We thank Melissa Dawes for professional language editing.

**Conflicts of Interest:** The authors declare no conflict of interest.
