Weather forecasting first and foremost is about observation

Weather forecasting first and foremost is about observation

Droughts, violent storms, heavy rains… predicting the weather is vital for farmers. A thorough understanding of the weather is only possible through good observation and forecasting. Meet Louis Bodin, one of France’s specialists in this field.

‘Predicting the development of the weather and the natural world depends first and foremost on good observation of the weather,’ explains Louis Bodin, TV weatherman and meteorological engineer. As Sencrop’s special guest at our webinar on 18th March, the stand-out star of French weather reporting was eagerly awaited by the farmers who were invited to participate in this online conference.

The weather forecast is ‘a passion’ for the man who presents the weather bulletins every week on a national TV channel. His decades of experience have taught him ‘humility in a field that has its limits’.

Farmers facing the challenges of the weather

It is not unusual to see fields ravaged by hail in summer, or sometimes by late frosts in early spring. The main topic for farmers wanting to protect their crops was: what is the best possible way to predict the weather in order to plan ahead effectively? The role of weather forecasting is crucial to this.

A variety of different models exist to predict the weather. We asked the participants the following question: ‘Do you know which weather forecasting models are behind weather apps and websites?’ Around 70% said ‘No’. This was no surprise for the meteorologist, who acknowledges that the general population ‘knows little about where the forecasts come from’. There are several big weather models around the world, such as the American GFS, the Swiss meteoblue, and their French counterparts ARPEGE and AROME. These models are all different and are not used in the same way day to day, and therefore do not produce the same forecasts. Even less so on a local scale.

But, as one of the participants asked: ‘Why are there so many differences between these models?’ This question came from Nicolas Ragot, a winegrower and keen weather-watcher. The reason is that they all work with different prediction formulae, known as algorithms. ‘They differ depending on the country,’ explains Louis Bodin. ‘One model might work better under a number of conditions,’ clarifies the weather specialist, and explains the influence of the oceans or the mountains in local forecasting. For France, on a national scale, several influences must be taken into account, which are ‘the influence of the Atlantic, the continent and the Mediterranean’. It all combines together. It’s much simpler to predict a good forecast for England than for France,’ he assures them. The graduate in meteorological engineering insists on the importance of taking these parameters into account. ‘It’s important to take your time and say to yourself: well, for my geographical location, for my area, this particular model is definitely going to perform a bit better than that other one.’ It’s possible to learn this with time, analysis and observation.

Is weather forecasting a reliable science?

‘No, not 100%,’ he acknowledges, admitting to ‘making mistakes on average one day out of 30’ when writing his weather bulletins. ‘The problem is that mistakes can sometimes end badly for farmers, for example. But that’s just how it is, unfortunately. We do try to keep errors to a minimum,’ insists the forecaster during the conference. The first stage remains observation first and foremost, with weather stations, aeroplanes and specialist satellites. Other tools are used to refine the forecasts, such as weather balloons sent up into the atmosphere to harvest data on temperature and humidity at high altitudes. ‘If we want to make a good forecast, we have to know what the weather is like all over the planet, because it all interacts,’ he adds. It’s an alchemy ‘between the observations and the power of the calculations, which allows the computers to produce forecast maps,’ the specialist explains.

A Sencrop study determines the reliability of weather models

‘The weather forecast is part of our daily lives; we look at it several times a day, but sometimes we rage at it a little. We would prefer it to be better,’ concedes Nicolas Ragot, winegrower. The keen weather-watcher acknowledges that he gambles on a combination of his own observations, thanks to his various weather stations, and the weather models. ‘In winegrowing, it’s essential to anticipate the weather. For example, when I see significant frosts coming,’ he emphasises. However, he is very cautions against ‘looking for a model that announces more fine weather than another’.

A recent study carried out by Sencrop last January details the quality of the forecasts given by these weather models. The report was created by comparing the data from Sencrop weather stations to the predictions of various models. ‘We gave a score for each criterion. For example, for temperature we checked the discrepancy between the prediction and the reality,’ comments Kévin Guilbert, head of the Sencrop app, who was in charge of the study. ‘It’s not that one model is better than another; it all depends on the area. We have noticed that the AROME model, which belongs to Météo France, does well at short-term temperature predictions for Champagne-Ardennes, but a little less well in the west of the country. The Dutch model, ‘Harmonie’, has an advantage for the central to western regions. For long-term forecasting, ‘there is a German model that’s ahead with regard to consistently reliable results over time,’ explains the lead author of the Sencrop study. The study covered 300 stations in Europe, comparing them with 10 prediction models. It confirmed that the reliability of a model depended on several criteria, such as the location of farms, the forecast period, seasonality, or again the temperature. And on the subject of these findings, the lover of wind and fog, Louis Bodin, confirms one thing: ‘All countries want to have observations and forecasts that are more and more complete and reliable.’ Could this be good news for the weather forecasting of tomorrow?

by Kévin Floury