A recent study signed by, among others, researchers of our Hydrology and Hydraulics Department, describes how innovative, high-resolution satellite data can be used in hydrological models to represent the water cycle, and monitor and predict extreme events. The work is the result of the ESA’s DTE Hydrology project, now in its ideal continuation with DTE Hydrology Evolution project
Not all areas of the planet have, or can be reached with, ground-based sensors that allow us to observe the dynamics of occurring events, such as floods. When present, these sensors provide spatially limited information. That’s why research is increasingly focusing on how satellites, with their broad look from above, can allow us to gather diverse information, of various kinds, about the characteristics of the elements present and the events taking place on the ground. But how do we integrate satellite information with the modeling tools we use to reconstruct, for example, the water cycle and to predict and monitor extreme events such as floods?
A study, recently published in Hydrology and Earth Systems Sciences, authored by an international team including researchers from our Hydrology and Hydraulics Department, shows how the use of innovative, high-resolution satellite data can be used for hydrologic modeling, and thus for reconstructing the water cycle and extreme events. The work is the result of the DTE Hydrology project, funded by ESA and completed in September 2021, which was aimed precisely at understanding the potential of satellites in this field.
Simone Gabellani, head of the Hydrology and Hydraulics Department and author of the study, tells us about the project and the results reported in the article.
High-resolution data
“The DTE Hydrology project is part of a strand of ESA research and technological development that aims to develop a Digital Twin Earth (DTE), a kind of interactive model capable of reproducing the dynamics of the planet by integrating satellite data with data from other sources, Big Data and artificial intelligence,” Gabellani says. “The work was divided into different sectors, for example Antarctic, Oceanic, Forest systems. As CIMA Research Foundation, we were involved in the activities concerning the hydrological sector: our goal was to understand how high-resolution satellite instruments can help us describe the water cycle and predict some related extreme events (floods and droughts, but also landslides), as well as for the monitoring and management of water resources”.
The research activities, reported in the new study, thus focused on innovative data that satellites can provide us: thanks to new sensors and algorithms, these instruments allow us to observe the Earth at a spatial scale that would not have been possible just a few years ago, estimating parameters such as evapotranspiration (i.e., water evaporating from soil and plants) at a resolution of one kilometer. In addition, because of their continuous orbit around the planet, satellites provide diffuse information, thus over a wide area, and not just punctual local information as ground-based sensors do.
The case study on Po River basin
The data that those involved in the study focused on included, in addition to evapotranspiration, snowpack height, precipitation and river discharge. As a case study, the research team used the Po River basin: as Italy’s main waterway, it is of great importance not only for the country’s agriculture and livestock but also for hydropower production, because some of Italy’s main dams are located along its course. Precisely because of its nationally significant role, the Po River basin is closely monitored with numerous ground-based sensors, which enabled the first fundamental step in validating the research results, allowing the data collected by satellites to be verified by comparing them with those recorded on the ground.
Once the validity of the satellite data had been confirmed, the research team wanted to see if, when integrated into the models, they allowed for an accurate reconstruction of river dynamics. “Some data, such as those concerning precipitation, can be used as inputs into the models, that is, entered to allow the model to perform calculations. Others cannot used directly as inputs: these are those, such as soil moisture, referred to as “state variables,” for which the scientific literature has already shown that they can be used with data assimilation algorithm, a methodology that allows us to integrate (or rather assimilate) data into the model by correcting the modeled state as we perform calculations. “In other words, this technique allows us to ‘correct’ the calculations as the information changes, following the evolution of the system over time,” Gabellani explains.
The research team conducted several experiments, varying the type of data provided to the hydrological model (represented by Continuum, a physical model developed by CIMA Research Foundation) to see how they influence the capability of the model in reproducing the flow rates actually observed in the Po. “The results show that precipitation is the variable that most influences the model result. This is not surprising, because it is the water that reaches the ground that most determines the flow rate of a river. Satellite estimates of precipitation are among those still often subject to uncertainty: however, our study shows that even these data yield very good results,” the researcher says. “What’s more, even in experiments conducted using only and exclusively satellite data bring the results comparable with those obtained by employing ground-based sensors”.
Next steps
In short, the information coming from the satellites shows promise for use in the hydrological models we use to monitor and predict events such as floods and droughts. Continuing research in this field, increasingly refining the quality of satellite data and its use in modeling, is an important support for studies, monitoring, and decision support in areas that are difficult to access on the ground and have a wide range. The continuing evolution of satellite instruments, which push their resolution to ever greater accuracy (in some cases, as low as a few hundred meters), opens up many possibilities in this perspective.
“We are now continuing work in this direction, both as a partner in the ideal continuation of DTE Hydrology, the DTE Hydrology Evolution project, and trying to integrate more and more information into the model,” Gabellani concludes. “This also includes information regarding, for example, the use of the water resource for hydropower generation and the irrigation network, so that we have a closer representation of the actual movements of water – a task that requires us to get very close to those who live and work on the ground, to gather their knowledge. And finally, we are also working on artificial intelligence, to understand how it can help us ingest data into numerical models”.