Weather alarms

When the skies open

Should we put this all down to climate change? Maybe yes, maybe no. What is certain is that during the past few years, Europe has been regularly struck by various meteorological catastrophes: violent storms, unprecedented flooding, prolonged heatwaves and huge forest fires. The list of victims is growing and compensation claims are scaring insurance companies. During this time, research has been advancing, with the objective of improving forecasting accuracy.

© Shutterstock © Shutterstock
Professor Volker Wulfmeyer wishes good luck to the Cops project at the first balloon launch. © Universität Hohenheim/Murat Professor Volker Wulfmeyer wishes good luck to the Cops project at the first balloon launch. © Universität Hohenheim/Murat
The Perth–Inverness line (Scotland) during the December 2006 floods. © Scottish Environment Protection Agency (SEPA) The Perth–Inverness line (Scotland) during the December 2006 floods. © Scottish Environment Protection Agency (SEPA)

The Mediterranean coastal regions are particularly vulnerable to heavy rains. The soil, left dry and impermeable by long periods of hot weather, is no longer able to absorb sudden and intense rainfall. The water rushes down to the bottom of the valleys, swelling rivers into torrents which carry away everything in their path. The human and material damage can be substantial. Between 1950 and 2000, some 2 200 people died in Spain as a result of this phenomenon, 815 of them trapped by rising water and dead within 3 hours. More recently, in 2002, floods in the French département of Gard cost the lives of 23 people while the bill for the material damage reached €1.2 billion.

To limit the impact of these flash floods, the Flash research programme, part of the 6th European Framework Programme, and with funding of €1.2 million from the Commission, has set itself the objective of improving forecasting of the sudden, intense downfalls which cause such damage, both in the short range – 1-2 days – and in the very short range – 2-3 hours. “We know that a correlation exists between heavy rainfall in thunderstorm zones and electrical activity within these same zones”, explains Colin Price, who coordinates the Flash programme at the University of Tel-Aviv (IL). “Several studies have demonstrated that, in a thunderstorm, increased electrical activity frequently corresponds to an increase in rain intensity.” The mechanisms of the relationship, however, remain unclear.

Twenty three floods as witnesses

Unlike rain clouds, which radars can pick up only up to a few hundred kilometres away, thunderstorms can be easily observed at much greater distances. This is because the lightning discharges emit very low frequency electrical electromagnetic waves (micro-waves), which travel several thousand kilometres and can therefore be picked up a long way off, in particular by satellites. Forecasting the formation of a storm zone is therefore possible, but knowing where and with what intensity the rain is going to fall is quite a different matter. It is here that the Flash researchers come in.

“We began our research”, Colin Price continues, “by analysing a posteriori a series of 23 flash floods following sudden cloudbursts in the Mediterranean basin, and for which we have full rainfall and electrical activity data. These data come from weather satellites and radars, but also from ground-level observations by weather stations. We then integrated these data into computer models that enable us to forecast heavy rainfall in the short and very short term. By taking each of the 23 flash floods and comparing the predictions given by these models with what in fact happened, we can assess the reliability of the algorithms used. The algorithms that work best with these past examples can then be tested and used in real time. In this way, people can in future be warned of imminent heavy rainfall.”

Once these models are up and running the Flash researchers will not simply pack their bags and go home. It is also important to have reliable models that can forecast how the rainwater will be distributed in the hydrological basins of each affected region. Depending on the geology, relief, level of urbanisation, impermeability of the soil and many other parameters, a heavy downfall can become torrential or not, devastating or not. “This is Flash's main challenge”, Colin Price is keen to stress. “Combining observation data in different formats and space-time scales, then integrating them into just as varied models is a major difficulty that we shall have to surmount in order to achieve the public utility goal which has been set for us, of being able to forewarn local populations of imminent flash flooding and avoiding as much damage as possible.”

Cops, a hyper-precise model

Change of scene. In June 2007, a scientific armada code-named Cops (Convective and Orographically-induced Precipitation Study) was deployed over a three-month period in a zone between the Vosges (FR) and the Black Forest (DE). Its task was to observe how thunderstorms form. Why here and why at this time of year? “Because the summer season is propitious for the formation of thunderstorms”, explains Evelyne Richard, research director at the Centre national de la recherche scientifique – CNRS (FR) and coordinator of the French part of Cops . “This tendency is reinforced by the semi-mountainous relief of the zone we studied. By choosing this region we were certain to have a sufficiently large number of case studies.”

But Cops is also part of a much broader project to introduce new digital local weather forecasting models, supplementing those of the ECMWF (European Centre for Medium- Range Weather Forecasting) (1) at Reading (UK), and replacing existing secondary models in Germany and France. “Over many years, the models have improved considerably in terms of forecasting temperature, cloud cover, wind, etc. But rain forecasting has advanced very little. We need to be able to correct this bias for the next models”, Evelyne Richard continues. “One perspective is to considerably improve the model’s precision by tightening the grid. We need to end up with 2.5-km grid squares and altitude layers of around 10 metres for those layers closest to the earth’s surface. Even if it is still too early to draw precise conclusions, we can already say that we have made progress, since certain models we are experimenting with have been able to forecast storms which were not predicted by the operating models.”

Ground-air operation

The high expectations placed on the project were certainly matched by the scale of resources deployed. Five advanced observation posts were laid out on an east–west axis, from the Vosges to the edge of the Black Forest, passing through the Rhine valley. All were fitted with the same cutting-edge equipment forscrutinising the sky and picking up all the potential components – temperature, wind, water vapour, particles – of “summer convective precipitations”, more commonly known as thunderstorms. Researchers were able to track the evolution of storm clouds from their formation, generally on the heights of the Vosges, through to their dissolution downstream from the Black Forest.

Radars and lidars (laser radars) incorporated into these ephemeral weather stations were backed by a network of 80 GPS stations across the study zone. Why GPSs? Because satellite signals beamed to these locating devices are slightly delayed when the atmospheric humidity level is high. This particularity can be used to calculate the amount of water vapour between the satellite and the GPS base.

To supplement these observations, Cops also used two mobile radars to compensate the gaps between the 80 GPS stations and the five weather stations. In the same spirit, teams criss-crossed the region releasing weather balloons. Finally, a fleet of no less than eight super-equipped aircraft flew over the study, each carrying radars and lidars for examining the atmosphere under the aircraft’s belly, each releasing large numbers of dropsondes.

“In three months, thanks to this hardware”, concludes Evelyne Richard, “Cops was able to gather a large quantity of data. These were introduced into test models which were adjusted as time went on. In the longer term we should have a reliable model, at least for our subject area, which is forecasting thunder storm precipitations in zones with very hilly relief patterns.”

Matthieu Lethé

  1. See article « Reading: the impossible weather equation


Read More

Alert in four colours

Thursday 8 November 2007. Panic stations at the KNMI Koninklijk Nederlands Meteorologisch Instituut, the Dutch national meteorological institute. Very high tides are expected that evening and the next day which could endanger the country's infrastructures and populations. The authorities are immediately warned to enable them to take appropriate measures. At the same time, the KNMI sends out a signal to Meteoalarm to put the country on maximum red alert.

The Internet site, launched on 23 March 2007 at the initiative of Eumetnet – a consortium of public weather forecasting services – is tasked with alerting web users to weather-related risks like floods, storms, heatwaves, fog, snowfall, high winds and forest fires. For this it uses a four-colour code – green, yellow, orange and red depending on the level of alert – enabling users to evaluate the situation at a glance, in each region of each participating country. “Our strength”, says technical director Michael Staudinger, “lies in having put together a harmonized European alert system on a single platform. Meteoalarm allows many people travelling abroad to find a single weather alert system in each country visited.”

With 280 million clicks since the site was opened, and with peaks of 12 million clicks a day during alerts, Meteoalarm serves both the general public and emergency services. Managed by the Austrian meteorological institute with funding of € 2 million, it receives regular information from national weather services – “some send their files every five minutes, others four times a day”, Michael Staudinger tells us. Future developments include “opening the system to other European countries or setting up an intranet service though which civil protection services can prepare for alerts up to five days ahead.”


To find out more