π Science and technology
How revolutionary AI and satellites are changing weather predictions

Better weather predictions with a new fleet of European satellites

Sébastien Léas, weather forecaster at Météo-France
On March 29th, 2023 |
3 min reading time
Sébastien Léas
weather forecaster at Météo-France
Key takeaways
  • In 2030, Third Generation Meteosats (TGM) will be launched to enable better identification of weather phenomena.
  • Their images will be twice as accurate and reliable than previous versions, being refreshed every 10 minutes.
  • MTGs can detect extreme weather events, which will become increasingly frequent due to climate change.
  • These satellites will use 100 times more data than the second generation.
  • They will be able to create multispectral images, condensing observations from different satellite channels.

Fore­cast­ers will be able to iden­ti­fy extreme weath­er events more quick­ly and effi­cient­ly, thanks to four new imag­ing satel­lites and two sounder satel­lites, sched­uled to be launched between 2022 and 2030. These Third Gen­er­a­tion Meteosats (or MTGs) will grad­u­al­ly replace the sec­ond-gen­er­a­tion satel­lites cur­rent­ly oper­at­ing in space. The first of these will be launched at the end of this year, to be oper­a­tional in 2023, fol­lowed by three sim­i­lar mod­els and two sounder satellites. 

With them, the images sent to Earth will be twice as accu­rate and reli­able: they will be refreshed every 10 min­utes (com­pared with 15 min­utes today), in oth­er words, in near real-time. They will also have a new light­ning detec­tion instru­ment (called the Light­ning Imager), nev­er before seen in Europe, which will be able to observe light­ning much more accu­rate­ly than cur­rent sys­tems; and oper­at­ing from the ground so unable to detect light­ning between clouds and those about to hit the ground. 

They will also be able to detect severe thun­der­storms and oth­er extreme weath­er events at an ear­ly stage. These types of events are like­ly to become more fre­quent in the future due to glob­al warm­ing. The new obser­va­tions will improve our knowl­edge of these events and allow us to warn peo­ple when necessary.

One hundred times more data 

The data from these satel­lites – one hun­dred times more than those obtained by the sec­ond-gen­er­a­tion satel­lites – will be used in new mod­els by the Euro­pean Organ­i­sa­tion for the Exploita­tion of Mete­o­ro­log­i­cal Satel­lites (EUMETSAT), the coor­di­na­tor of the Euro­pean weath­er satel­lite net­work. In France, the Cen­tre de météorolo­gie spa­tiale in Lan­nion, Brit­tany, is respon­si­ble for pro­cess­ing the data. The advan­tage of this site is that it has good recep­tion, as it is not pol­lut­ed by oth­er satel­lite receivers found in larg­er cities, such as Toulouse, which is the con­trol cen­tre of Météo-France.

The aim is to make the best pos­si­ble fore­casts so that we can pre­vent and secure peo­ple and property.

The Sec­ond Gen­er­a­tion Meteosat (SGM) has an imag­ing radiome­ter that oper­ates in the vis­i­ble and infrared parts of the elec­tro­mag­net­ic spec­trum. This imager observes the Earth in 12 dif­fer­ent chan­nels with a res­o­lu­tion of 1 km for the High-Res­o­lu­tion Vis­i­ble chan­nel and 3 km for the oth­er chan­nels. Com­pared to the MSG, we have even more chan­nels with the MTG and more scans, so bet­ter accu­ra­cy. This infor­ma­tion will be use­ful on a dai­ly basis, espe­cial­ly in high-risk phenomena.

After the launch, there will be a test phase fol­lowed by an ini­tial use phase where fore­cast­ers will learn to use the new instru­ments. They should have increas­ing­ly rich image and detec­tion qual­i­ty in sev­er­al chan­nels, reach­ing a very high res­o­lu­tion of about 500 metres for the chan­nel oper­at­ing in the vis­i­ble spec­trum. The aim is to make the best pos­si­ble pre­dic­tions to be able to warn and secure peo­ple and property. 

Multi-spectrum synthetic images 

The cur­rent Meteosat satel­lites pro­duce com­pos­ite and colour images, and the third gen­er­a­tion will do the same. The goal for the sec­ond and third gen­er­a­tions is to cre­ate mul­ti-spec­trum syn­thet­ic images by ‘con­dens­ing’ obser­va­tions from dif­fer­ent satel­lite channels. 

There are sev­er­al types of satel­lites: first­ly, ‘geo­sta­tion­ary’ satel­lites, which always observe the same place on Earth and are locat­ed at an alti­tude of more than 30,000 km. They rotate at the same speed as the Earth and allow us to car­ry out weath­er ani­ma­tions. In con­trast, the so-called ‘non-sta­tion­ary’ satel­lites revolve around the Earth and do not see the same band of tra­jec­to­ry. These satel­lites are locat­ed at a much low­er alti­tude, only 800 km, which makes it pos­si­ble to obtain much more pre­cise images, espe­cial­ly of low clouds or fog. 

By using the dif­fer­ent satel­lite chan­nels, we can make what is called an ini­tial state of the atmos­phere. This will allow us to cor­re­late what we observe with what we fore­cast. To make a good fore­cast, we need to know what is hap­pen­ing now but also what hap­pened a few days ago, for exam­ple, a lit­tle fur­ther out in the Atlantic. This will allow us to observe the evo­lu­tion of the cloud mass­es and to make a good rep­re­sen­ta­tion of them.

The goal is to cre­ate mul­ti-spec­trum syn­thet­ic images by con­dens­ing obser­va­tions from dif­fer­ent satel­lite channels.

The result­ing images are in black and white gra­da­tions, a kind of pho­to­graph­ic rep­re­sen­ta­tion of the cloud reflec­tiv­i­ty. The bright white colours rep­re­sent clouds that are gen­er­al­ly very thick, as they strong­ly reflect sun­light. Small­er clouds are grey­er, even a lit­tle dark­er, and are often low clouds that are laden with rain and there­fore have a low reflec­tiv­i­ty, as they absorb more light.

At night, how­ev­er, we use the infrared chan­nels, which take the tem­per­a­ture of the first lay­er of clouds encoun­tered. The high­er the cloud, the low­er the tem­per­a­ture. And the low­er the tem­per­a­ture, the lighter the colour. This is why high clouds appear white in the infrared. For a short-term fore­cast, an ani­ma­tion of sev­er­al images should be con­sult­ed. How­ev­er, analy­ses are more dif­fi­cult at cer­tain times of the year – autumn, for exam­ple, when we some­times find low clouds that have the same tem­per­a­ture as the ground, mak­ing them dif­fi­cult to distinguish. 

Interview by Isabelle Dumé




Our world explained with science. Every week, in your inbox.

Get the newsletter