).Remote Sens. 2021, 13, 4025. https://doi.org/10.3390/rshttps://www.mdpi.com/journal/remotesensingRemote
).Remote Sens. 2021, 13, 4025. https://doi.org/10.3390/rshttps://www.mdpi.com/journal/remotesensingRemote Sens. 2021, 13,two ofchange, all-natural catastrophic events (i.e., wildfire), and anthropogenic activities, for instance intense irrigation practices, water drainage, groundwater extraction, and replacement by urban and agricultural landscapes [13]. Thus, it can be very important to acquire precise, trustworthy, and up-to-date facts regarding the distinctive traits of wetlands (i.e., extent, kind, overall health, and status). Traditionally, wetland mapping was carried out by collecting airborne photographs and in situ information by means of intensive field surveys [14,15]. While these methods have been quite correct, they had been resource-intensive and practically infeasible for large-scale research with frequent data collection necessities. Consequently, advanced Remote Sensing (RS) tactics have been proposed for wetland mapping and monitoring [2,168]. RS systems supply frequent Earth Observation (EO) datasets with diverse traits and broad region coverage, creating them desirable to map and monitor wetlands’ dynamics from nearby to worldwide scales via time [2,19,20]. Having said that, it need to be noted that the possibility of acquiring dependable facts about wetlands employing RS data will not obviate the necessity of collecting in situ information, and their incorporation shall give far more profound outcomes. Passive and active RS systems capture EO data in unique components from the electromagnetic spectrum. Within this regard, aerial [213], multispectral [18,247], Synthetic Aperture Radar (SAR) [281], hyperspectral [20,32], Digital Elevation Model (DEM) [336], and Light Detection and Ranging (LiDAR) point cloud datasets [368] have been extensively applied separately or in conjunctions for wetland mapping. Because each and every of those information sources acquire EO data in unique components of the electromagnetic spectrum, they present diverse information about the spectral and physical characteristics of wetlands [39]. Furthermore, deployment of those sensors on airborne, spaceborne, and Unmanned Aerial Car (UAV) platforms allows recording EO data over wetlands with unique spatial resolutions and coverages. Finally, the integration of RS information with machine learning algorithms delivers a fantastic opportunity to totally exploit RS data for precise wetland mapping and monitoring tasks [40,41]. Machine finding out algorithms let extracting and interpreting RS information automatically and robustly to map wetlands and derive relevant information and facts regarding the wetlands’ condition. As an example, Random Forest (RF) [425], Support Vector Machine (SVM) [469], Maximum Likelihood (ML) [503], Classification and Regression Tree (CART) [35,36], and Deep Finding out (DL) [21,27,40,54] algorithms happen to be DSP Crosslinker medchemexpress implemented to generate highquality wetland maps. Within this regard, each pixel-based and object-based approaches happen to be applied to exploit the most delicate doable info about wetlands by integrating RS information and machine mastering algorithms [552]. In addition, studies [21,40,41,47,48,63] have been also committed to assessing the functionality of machine learning algorithms for precise wetland mapping and monitoring to elucidate the path for other interested researchers all about the globe. International wetland extents had been predicted to become from roughly 7.1 million km2 to 26.six million km2 [64] and 25 of globally documented wetlands have already been recorded over Canada, covering about 14 on the total Canadian terrestrial surface [.