Mapping of shore area wetlands in Lake Tana: Using Multi-Source Datasets, Sentinel-1A SAR, and PCI Geomatica Software

Abstract Shore area wetlands are important ecotones for biodiversity, spawning and refuge habitats, high primary productivity, and significant material exchange between aquatic and terrestrial ecosystems. The distribution, size, and dynamic changes of lacustrine wetlands must be understood in order to effectively manage and maintain these resources. This work aimed to develop accurate shore area wetland mapping with multi-temporal, multi-source data, wetland indicators, and radar imagery of Sentinel 1 using Geomatica software, ground truth validation, and manual techniques. The topographic position (DEM and derivatives, as well as Lake Bathymetry) and hydric soil wetland indicators were mapped using different multisource datasets and Arc GIS software. Hydrophytic vegetation and wetland hydrology were mapped using publicly available Sentinel-1 SAR data and Geomatica Banff software packages. For shore area wetland mapping, the analytical hierarchy process and weighted overlay method were applied. The topography position wetland indicator map covered about 55,363.53 ha, while hydric soils (Gleysols, Vertisols, and Fluvisols) covered about 55,151 ha around Lake Tan within a 3 km buffer. The hydrology wetland indicator map of only permanently inundated covered about 591,311.43 ha, while including temporarily inundated covered about 607,052.48 ha. Hydrophytic vegetation around Lake Tana, including invasive water hyacinth, covered about 74,771.86 ha. The overall land use/cover classification accuracy and overall Kappa statistic were 78.68% and 0.70, respectively. Based on Class Accuracy Statistics results of Producer Accuracy (PA), User Accuracy (UA), and Kappa Statistic (KS), water bodies (PA = 94.11%, UA = 99.12, and KS = 0.98), cultivated land (PA = 77.20%, UA = 81.04%, and KS = 0.780), and hydrophytic vegetation (PA = 69.69%, UA = 59.77%, and KS = 0.51) were classified with the highest accuracies. Shrub-land (PA = 45.00%, UA = 29.03%, and KS = 0.27), and Plantation of Eucalyptus (PA = 0.00%, UA = 0.00%, and KS = -0.02) had the lowest PA, UA, and KS. Shore area wetlands were generally found around Lake Tana, about 3 km to the terrestrial part, and covered about 26,663.24 ha. The combination of multi-temporal, multi-source data, wetland indicators, radar imagery of Sentinel 1 using Geomatica software, ground truth validation, and manual techniques provided information on the spatial distribution of shore area wetlands in Lake Tana. In order to properly manage and preserve these resources, the distribution and extent of shore area wetlands produced by this study will be a crucial point of reference for next research.