Akkoord cao universiteiten op handen

Bestuurslid Marieke van den Berg van Abvakabo FNV bevestigt geruchten over een akkoord dat op handen zou zijn. De onderhandelingen gaan volgens haar de goede kant op, maar ze laat in het midden of de door de bonden gewraakte nullijn voor de salarissen verlaten is.

“We zijn er nog niet”, tempert VSNU-onderhandelaar Hugo Levie de verwachtingen. “Er moet nog wat water onder de brug door.”

It may seem logical for the water boards that control water systems, like river basins or polder areas, to use weather forecasts to help manage their systems. But in practice, they often don’t. “In some systems, it just isn’t necessary. It can for instance suffice to use measurements of the upstream water levels in a river system. But in systems where quick action is needed in the case of extreme weather events, forecasting data can be useful”, says Schalk Jan van Andel. For his PhD research at the Unesco-IHE Institute for Water Education, Van Andel developed a method to help water boards make management decisions based on weather forecasting and hydrological predictions – a tactic called anticipatory water management. His thesis defence was on November 3.  

Van Andel didn’t just look at normal weather forecasts: he focused on so-called ensemble forecasts. “The problem with weather forecasts is that, because weather is a chaotic system, there is always a level of uncertainty involved. So you might for instance take water control measurements that are unnecessary. Ensemble forecasts make it easier to work with the uncertainty factor”, he explains.  Basically, ensemble forecasts are composed of a large number of predictions that were made using slight variations in weather values. The less variation there is within this set of predictions, the greater the certainty that the weather will be like the ensemble forecast predicts. Another advantage of ensemble forecasts is that they allow you to look further ahead than normal forecasts.

Van Andel designed a model that incorporates ensemble forecasts, a model for the hydrology of a water system, the different management decisions that can be taken to control the water system, and the effects of these decisions. As a test case he modelled the Rijnland Water Board, a polder system in the west of the Netherlands. He then used this model to test many different decision strategies in different situations. Van Andel: “Due to advances in hydro informatics, we’re now able to model operational water management quite well. And it’s a lot faster, as well as safer, to test the effect of different decision strategies in a computer model than it is to test them in practice.”
The running of the model gave Van Andel a set of decision rules on how to deal with weather forecasts for the Rijnland system, which he then validated and fine-tuned by feeding the model with historical data of the water system, as well as historical weather forecasts, and applying the rules to this data set. It turned out this computer-designed anticipatory water management was quite beneficial compared to the traditional re-active way of management. By using the ensemble forecasts and the optimised set of decision rules, it was possible to anticipate upcoming events three days in advance, and this led to a 30 percent reduction of flood damage costs in the eight-year period that Van Andel modelled.

An important focus of the Unesco-IHE institute, where Van Andel conducted his research, is to improve water management in developing countries. Van Andel therefore also modelled the Upper Blue Nile water system in Ethiopia, in order to show that his approach can also be used in developing countries. “Ensemble weather forecasts scales are freely available on the internet. Also, the software for making hydrological models is becoming more readily available. So there is not much preventing other countries around the world from developing anticipatory water management”, Van Andel explains. He does however stress that, since every water system is different, a new set of decision rules must be made and validated for each water system. But this doesn’t have to be costly, or involve lots of work, because the computer can do most of it. 
Van Andel still has the model that incorporates ensemble forecasts and the automated decision rules for the Rijnland water system running on his work computer, in real time. “The Rijnland water board is very interested in it”, he says. “But they cannot adopt it just yet. Before they can do so, the model and its operational reliability should be further tested.”