On 2 July, DigiShape organised a working session of the NL AIC Water and Climate Working Group to explore how AI can help with water issues. One of the concepts discussed was VARAA, which aims to use AI and satellite data to create accurate, up-to-date and useful information for water management. We spoke to Joost Heijkers, hydrologist at Hoogheemraadschap De Stichtse Rijnlanden and creator of the VARAA concept.
What does VARAA entail and how did the idea originate?
“VARAA” stands for: Visualization, Acceleration, Representativeness, Accuracy, Actuality. The concept arose from a need to make the combination of AI and satellite data practical for water managers by generating useful hydro information that meets the VARAA criteria: Visually Appealing, Quickly Created & Delivered, Representative, Accurate and Current. While AI and satellite data have great potential to improve our hydrologic system understanding, actual combined application and use lag behind. The challenge we face is to combine AI and satellite data into useful hydro information that enhances the action perspective of water managers.”
How can VARAA bring progress?
‘Innovations often fail at the proof-of-concept stage and then don’t reach the shop floor. VARAA aims to bridge that gap by turning AI and satellite data into directly useful information for water managers. Although there is a lot of satellite data available, it is not yet sufficiently utilised in the water sector. VARAA makes this data available faster, more accurate and more representative of the areas where decisions need to be made.
AI can be used both on Earth and in space; to create information from raw data. For example, a company like Cosine has built their AI algorithms into satcubes: mini satellites shot into space by the dozens at a time. These satcubes are a welcome addition to ‘traditional’ satellites. Through smart combinations, we can provide information on entire areas, such as the whole of the Netherlands, but also a specific city or nature reserve. This makes for faster and more informed decisions.’
How do you see the role of AI and satellites in the future of water management?
‘I think AI and satellite data will greatly enrich the world of traditional hydrological monitoring within a decade and perhaps partly take over in the longer term, the latter especially from efficiency objectives. The problem with current satellite data is that it is often not directly usable for the Dutch situation. We can solve this through VARAA, which scales down data with AI to a usable space-time scale for use.’
What is your dream scenario?
‘I have been active within the STOWA SAT-WATER Programme since 2015, which aims to promote the use of satellite data and derived information among Dutch water managers. Together with Hans van Leeuwen (STOWA Programme Manager), Het Waterschaphuis and other water managers, we try to translate this programme into operational information for the shop floor. This has led to projects such as SATDATA for evaporation data, LIBV for soil moisture data and NL Change Detection for changes around water systems and flood defences. Our next goal is to enable satellite-based estimates of subsidence and water quality.
With our tenth anniversary in sight, we want to develop a new vision: the integration of satellites, AI and ‘classical’ hydrological models via the Dutch Hydrological Instrumentarium (NHI). This combination can accelerate the water world by providing more and better hydro information in line with the VARAA criteria.
Imagine, for instance, that we can visualise the future groundwater status in an area like the Utrechtse Heuvelrug for the coming year (with seasonal weather forecasts). This could help water managers make well-informed decisions much earlier and adjust water management strategies to prevent drought or flooding. This is one of the applications we want to facilitate with VARAA, although there is still a lot of work to be done to make it happen.’
What do you think are the next steps to take VARAA forward?
‘The first step is to develop concrete use cases that show that AI and satellite data combined can provide accurate hydro information. Perhaps we can run pilots with the IJsselmeer, the Utrechtse Heuvelrug or Dijkring 15 as pilot areas. Here, it is important that we involve end-users (water managers), from the start. If they are not convinced of the information value, it will never get off the ground.
This is why communication is also so important; we have to make sure that we not only achieve good results technically, but also that those results reach the policymakers and administrators who have the resources to implement this kind of innovation on a large scale.’
Although VARAA is still at a conceptual stage, it has the potential to play an important role in the future of water management. In the coming months, the Dutch AI Coalition’s Water and Climate working group will explore, in consultation with the SAT-WATER consortium, how this idea can be further developed and implemented.
