A team of researchers led by IGB scientist Sami Domisch, have developed the highest resolution map of the world’s river systems ever produced. The map provides a basis for the detailed study of what characterizes riverine habitats and how they are interconnected. The map is based on the “Hydrography90m” data set.
“We took a high-resolution elevation model of the earth and used open source software to extract the river network from it. Unlike other previous data sets, Hydrography90m also maps short and very short arms of flowing waters,” Sami Domisch said. The shortest unit is 90 meters long. Since small rivers account for the largest proportion of the global river network (around 70%), they play a significant role in riverine biodiversity.
For the study, they use data from 30,000 gaging stations worldwide where quantities of water have been collected in defined river sections for years. In addition, the researchers have access to comprehensive data on a wide range of environmental factors such as temperature, precipitation, land use, soil properties and slope. In the model, these parameters are related to the amount of water measured in each case.
“In this connection, we work with machine learning. This means that with every new data set, our model gets better and better at recognizing which parameter variables are related to which water volumes,” said lead author of the study, Giuseppe Amatulli. If the model works, it can be applied to all river sections worldwide, even if there is no gaging station. In this case, the model determines the probable amount of water in the river, from the environmental parameters available for the entire area.
To validate the model, the researchers train it with 70% of the existing water quantity data sets. The model is then tasked with determining the quantities of water from the environmental parameters of the remaining 30%.