A Call for Unified Definitions and Classifications for Saltmarshes for Accurate Mapping of Extent: A Case Study From Sri Lanka

Abstract Saltmarshes are increasingly recognised for their ecological value. Their diverse nature, occurrence amongst complex ecological mosaics, and occurrence in both tidal-influenced and isolated inland contexts demand precise delineations and classifications for effective protection. This study classifies tropical saltmarshes in the Indian Ocean region through a literature review to identify the ambiguities. Vegetation based mapping using three approaches (object-based image classification, pixel-based image classification, and satellite embedding dataset with supervised classification) was tested using field survey data collected between 2024 and 2025, covering half of Sri Lanka’s coastline. The review revealed a common conflation of inland and coastal saltmarshes, leading to inconsistent extent estimates. Object-based image classification, pixel-based image classification, and satellite embedding dataset with supervised classification yielded saltmarsh extents of 16,406.74 ha, 15,995.50 ha, and 40,410.12 ha, respectively. Field surveys were carried out, focusing on succulent dominated saltmarshes validating 15,807 ha. Object and pixel-based image classification results closely aligned with this data, while satellite embedding dataset with supervised classification overestimated saltmarsh extent. Object-based image classification and pixel-based image classification estimated a coastal saltmarsh coverage of less than 0.25% of Sri Lanka, highlighting the need for greater protection of these ecosystems from further loss. More training data could reduce overestimation by the satellite embedding dataset with supervised classification. The study highlights the need for better definitions and classifications of coastal saltmarshes and urges the need to integrate vegetation, geomorphology, hydrology and climatic data to ensure coastal saltmarshes are segregated from saline inland saltmarshes.