Modeling Water Hyacinth (Pontederia crassipes) Distribution Using Geospatial Techniques: The Case of Lake Tana, Ethiopia

No Thumbnail Available



Journal Title

Journal ISSN

Volume Title


Addis Abeba Universty


Water hyacinth, an invasive aquatic plant, poses serious environmental and socioeconomic challenges. Remote sensing is essential to understand and predict how species are distributed in different habitats and environmental conditions and helps with monitoring and management activities. This study is intended to model the distribution and detection of the spatiotemporal dynamics of water hyacinth via four machine-learning models in Lake Tana, Ethiopia. The study employs 11 variables for modeling and 16 variables for spatiotemporal dynamics obtained from Sentinel-1 SAR bands, Sentinel-2A bands and indices, and bioclimate data sources. The models used 458 presence and 458 randomly generated pseudoabsence as response variables and tenfold bootstrap sampling. The models were evaluated using the area under the curve (AUC), receiver operator curve (ROC), true skill statistics (TSS), coefficient of rank correlation (COR), sensitivity, specificity, and Kappa coefficient, while the spatiotemporal distribution between 2016 and 2022 was evaluated using the overall accuracy and kappa coefficient. The findings demonstrate that the random forest model outperforms the other models, with AUC values of 0.93 and 0.95, TSS values of 0.77 and 0.82, and kappa values of 0.76 and 0.82 in the wet and dry seasons, respectively. B12 (16.3% and 19.7%), NDWI (14.7% and 12.4%), mean annual temperature (13.4% and 14.2%), and B5 (11.4% and 12.4%) were the most relevant variables during the wet and dry seasons, respectively, while B3, B5, B11, B12, VH, elevation, NDAVI and NDWI were the most relevant features in the spatiotemporal detection. According to the model prediction result, water hyacinths have the highest coverage during the wet season. The spatial coverage was 686.5 and 650.4 ha in 2016 and 1436.5 and 1216.5 ha in 2022 in the wet and dry seasons, respectively. Study results showed manual removal and machine harvesting were used to manage water hyacinth. The research concludes that the integration of Sentinel image indices and bands with bioclimatic variables is essential in the modeling and detection of spatiotemporal dynamics. The research recommends that geospatial technology helps in the regular assessments and timely detection of water hyacinths as a response to new infestations and prompt management actions.



Water hyacinth, remote sensing, machine learning, modeling, spatiotemporal dynamics, Lake Tana.