Measurement and forecast are two different things
On each river page, today’s flow is a measurement taken by the Public Service of Wallonia (SPW) at a station. For the following days, the “Forecast flow” is the result of a calculation. Only the measurement counts when deciding whether to go; the forecast helps you pick a day and avoid a wasted trip.
What the calculation uses
The model invents no data. It combines official information, stored and re-read automatically:
- the flow measured by the SPW at the river’s station, with the latest value, the hourly flows and the daily flows;
- the rain that actually fell on the catchment, recorded by several SPW rain gauges inside the catchment or on its edge, and counted over the same midnight-to-midnight day as the daily flow;
- air temperature, which drives evaporation: in summer, the soil absorbs most of a shower;
- the rain forecast by the weather service for the catchment, which is corrected for its usual bias after being compared with the rain gauges every day.
The model: a catchment that fills and empties
The calculation is based on GR4J, a reference rainfall-runoff model designed by the French research institute INRAE and widely proven. It describes three simple ideas.
- The soil holds back first. Rain fills the soil, less efficiently the wetter it already is. Evaporation empties it, a lot in summer and little in winter. The same shower therefore barely moves a river after a dry summer and makes it rise sharply in autumn.
- The water takes time to arrive. What the soil does not keep reaches the river over one to three days, with a delay specific to each catchment. A small, steep catchment reacts faster than a large one. This is why a forecast can keep rising for a day or two after the rain.
- The recession stretches out. Without rain, the river drains quickly right after a flood, then more and more slowly.
Each river has its own set of parameters, fitted to its past measurements. A model of this kind stays readable: you know why it gives a given result and, most of the time, where its error comes from.
Starting again from the measurement
The model does not run in a vacuum. Its state is reset on the measured daily flow, then, at the time of calculation, the flow recorded at that moment is compared with what the model would have given. Any gap is carried over to the following days, fading over time. Each night, a program re-evaluates each catchment’s parameters with the most recent measurements and keeps a new setting only if it improves past forecasts.
Horizon and translation into status
The river pages show the forecast day by day over the next five days. A sufficient flow does not guarantee that the river is open: the official rule judges the minimum on the average of the last 72 hours, which the calculation estimates for each future day. When a temporary ban is in force, the days concerned stay displayed as closed, even if the forecast flow is suitable. The status for the day is always the one published by the SPW. The rules are summarised in navigable flows in Wallonia.
The “best day” picks a day only if it meets three conditions: a forecast flow within the legal range, an official status expected to be open and a score of at least 60 out of 100, which takes into account the weather and the gap from the river’s ideal range. If no day qualifies, none is suggested. The rivers concerned are the Semois, the Lesse, the Ourthe, the Amblève, the Viroin and the Salm, which you can check from their pages, for example the Semois reading.
What the forecast does not guarantee
- The weather forecast can be wrong. The most frequent error comes from the forecast rain: a heavier or lighter shower than predicted changes everything else.
- The calculation is daily. It does not see very localised storms or the hourly peak of a flood.
- Structures escape the model. Dam or sluice operations change the flow with no link to rain.
- The Amblève is less predictable. The dams on the Warche upstream change its flow independently of rainfall.
- Reliability drops with lead time. It is best at one or two days; beyond that, read the forecast as a trend.
This method does not predict your skill level, the water temperature or the local conditions on your stretch. It does not replace the SPW decision on whether the river is open.
Reading the forecast well
Use it to compare the days of the week and to call a trip off when the trend is clearly unfavourable. On the morning of departure, check the measured flow and the official status, then look at the day’s weather. The site applies the same approach to all its rivers, and each page shows whether the forecast is an estimate or a measurement.