WunderBlog Archive » Weather Extremes

Category 6 has moved! See the latest from Dr. Jeff Masters and Bob Henson here.

Another Record Rainfall in Southern France

By: Christopher C. Burt, 8:12 PM GMT on September 30, 2014

Another Record Rainfall in Southern France

It is hard to believe that another rainstorm of equal intensity to that which I blogged about just 11 days ago has again struck the Languedoc Region of Southern France. This time the focus of the storm was centered over the city of Montpellier, Herault District, near the Mediterranean Coast.



Rue Broussonnet in Montpellier provided a new definition of ‘car pooling’ during the record-setting deluge Monday afternoon, September 29th when almost 10” of rain fell in three hours. Photo by Benoit from Tournemire.

Montpellier, a city of some 240,000, was deluged with an all-time 24-hour record 299 mm (11.77”) of rainfall between 8 a.m September 29th and 8 a.m. (local time) September 30th. This is the equivalent of almost four times the average monthly precipitation for September in the city. Its previous 24-hour rainfall record for September was 187 mm (7.36”) on September 22, 2003 and its all-time record 205 mm (8.07" on October 25, 1979. Some private weather stations in Montpellier reported totals of up to 325 mm (12.80”) according to a German weather forum monitored by Michael Theusner of Klimahaus in Bremerhaven, Germany.

Of the official 299 mm total in Montpellier, an amazing 184 mm (7.24”) of this fell in just two hours between 3 p.m. and 5 p.m. on Monday (September 29th) and 252 mm (9.92”) in three hours from 3 p.m. to 6 p.m. Here is the rainfall by time period beginning the morning of September 29th:



8 am-2 p.m.: 13 mm (0.51”)

2 p.m.-3 p.m.: 17 mm (0.67”)

3 p.m.-4 p.m.: 91 mm (3.58”)

4 p.m.-5 p.m.: 93 mm (3.66”)

5 p.m.-6 p.m.: 68 mm (2.68”)

6 p.m.-7 p.m.: 5 mm (0.20”)

7 p.m.-8 p.m.: 3 mm (0.12”)

8 p.m.-8 a.m. (Sept. 30): 10 mm (0.39”)

Graph and table of hourly (local time) rainfall at Montpellier Airport from Monday morning to Tuesday morning, September 29-30. (Note: the above totals amount to 300 mm instead of the official 299 mm because of rounding of figures. This also applies to totals expressed in inches). Source: Meteo France.

The two-hour total would be a new French national record for rainfall intensity, surpassing the 180 mm (7.09”) measured at Saint-Gervais-sur-Mare just 13 days ago on September 16th! Prior to this month’s extreme rainfalls, the previous greatest two-hour rainfall total observed in France was 178.4 mm (7.02”) at Solenzara on October 26, 1979. However, it should be noted that records in France for short-duration rainfalls only go back to the 1960s or 1970s and do not include every meteorological site in the country. Nevertheless, it is pretty shocking that two such amazing rain events have occurred in the same region over just a two-week period. Fortunately, unlike the September 16-18 event (when four died), it appears that, so far, no known fatalities have occurred with the storm in Montpellier.

The rainfall was concentrated directly over Montpellier as this map of accumulations illustrates:



Map of accumulated rainfall in millimeters for the 48-hour period ending at 6 a.m. UTC September 30th. Montpellier is located in that bull’s eye of 250 mm+ accumulations. The white area below Montpellier is the Mediterranean Sea. Map from Meteo France.

Other rainfall reports from the region were impressive although not close to that of Montpellier:

160 mm (6.30”) at Pezenas

156 mm (6.14”) at Beziers

153 mm (6.02”) at Prades-Le-Lez

124 mm (4.88”) at Narbonne

88 mm (3.46”) at Perpignan



Map of Languedoc-Roussillon region in southwestern France with location of Montpellier and some of the other cities mentioned above.

The storm brought this September’s total rainfall amount to 350 mm (13.78”), which is also a new monthly record (previous record being 293.2 mm (11.54”) in September 2003.

The cause of the intense rainfalls in both cases of September 16-18 and September 29 is an atmospheric set-up that is typical during the autumn in this part of the Mediterranean region (including Spain and Italy): warm, humid air flowing off the Mediterranean Sea collides with cooler dry air emanating from the Alpine region while a cold upper-air low rests over the area. Most of the greatest rainfall events in Spain, France, and Italy have occurred during September and October when this type of scenario is in play.

Meanwhile, the U.K. has wrapped up its driest September on record. More about that later.

KUDOS: Thanks to Michael Theusner of Klimahaus, Bremerhaven, Germany and Maximiliano Herrera for bringing this to my attention.

Christopher C. Burt
Weather Historian

Extreme Weather Precipitation Records

The views of the author are his/her own and do not necessarily represent the position of The Weather Company or its parent, IBM.

Reader Comments

As with the last record breaking floods in France i commented on your last blog that invest 90L (~September 4th 2014 off Africa's NW coast,  traveled across the Atlantic turning near Bermuda back eastward towards SW Europe attached to front and added (tropical like) precipitation in causing the last floods.

See this clip/animation of IR satellite imagery of 90L & 91L (Edouard). The animation ends just as 90L remnants blended into the Front that caused Frances 1st severe flooding of Sept 2014, clip here::
TITLED:: Dancing with the Dangerous clouds http://www.youtube.com/watch?v=JrUQq1bz2MQ&li st=UUZLKQ_XbXc7ELgp6NMumViA


Now 2014 Hurricane Edouard took a similar path as it began as Invest 91L BUT as Edouard and later Remnants of Edouard made a hard "right" / SSE to Southward turn just before the Azores, remnants of Edouard (to me) split into 2 . The higher clouds went eastward blending with a trough and eventually the trough passed over Puerto Rico & now (Sept 30th 2014) back northward over the eastern most Bahamian islands (i'm watching for TS genesis there & GoMx during the next 3 days). 

See Edouard's imagery tracks::

The 2 directions i state Edouards split into are TITLED::
uncoupled Remnants of 2014 Edouard 1 http://www.youtube.com/watch?v=nZCbMKJUEPI&li st=UUZLKQ_XbXc7ELgp6NMumViA
uncoupled Remnants of 2014 Edouard 2 http://www.youtube.com/watch?v=9ahGhgz02sA&li st=UUZLKQ_XbXc7ELgp6NMumViA
 
A few more days included as to Edouards remnants blending with trough
here TITLED:: Edouard to X http://www.youtube.com/watch?v=qksCITJFYxw&li st=UUZLKQ_XbXc7ELgp6NMumViA

The lower clouds actually had low level spin about 5 days ago and were moving seemingly ENE as to enter the Mediterranean sea, that's when my compu'r shorted, so i have no satellite imagery of those dates.

Since these are the rainy months for this area i'm surprised that being there where 2 tropical "plumes" heading in their direction that flooding was a surprise, NOT the magnitude (that was surprising) but that tropical moisture was blending with an already wet setup.
UK on course for driest September since records began in 1910
This month is set to be the driest September since records began in 1910, according to Met Office figures.......................... This September follows on from the eighth wettest August on record and comes in a generally very wet year - this January to August is the wettest such period in the records, mainly as a result of the very wet start to the year and the wettest winter on record, meaning that water levels are not a concern at this point.

Link
I'm curious about the expected frequency of these events. I.e. this the second ??? year event in eleven days?
Meteo France said the Montpellier rainstorm was a once in a 100-year event for the city. But what the return period for having two such rainstorms within two weeks in the same district is an open question.

Quoting 3. LowerCal:

I'm curious about the expected frequency of these events. I.e. this the second ??? year event in eleven days?
Lebanese mountains saw a great snowstorm ,the earliest in decades (in memory lives). The snowline was at 2700m and above.
At 3000m the accumulation was very deep. Snowflakes were recorded even slightly below 2500m.
There are no cases like this in the past decades at least in the first half of October.
Vladivostok in Russia also witnessed its earliest snowfall ever (29 September), the previous earliest was on 4 October 1963.
Quoting 4. weatherhistorian:

Meteo France said the Montpellier rainstorm was a once in a 100-year event for the city. But what the return period for having two such rainstorms within two weeks in the same district is an open question.



As in, conditions conducive to such an event may persist or recur over some period of time?
LowerCal there've been some snowfall records set in very quick succession in some US cities. Minneapolis set their all time record 24 hour snow accumulation on January 20-21, 1982, (17.4") and then set a new record on January 22-23, 1982 of 20.0". The first snowstorm ended 6 p.m. January 21st.. The second snowstorm began at 2 a.m. January 22nd.

A similar, although not as quite closely paired double record was set in Boston in 1978. On January 20-21st, 1978, there was 21.4" of snow in Boston, which set an all time record record, until February 6-7th 1978, the "Great Blizzard". Before January 1978 there hadn't been an official snow total of 20" in 48 hours from one snowstorm in Boston.

Between these two storms was the "White Hurricane" of January 25-26th, 1978 which brought a pressure of 28.28" to Cleveland and record snows to parts of the Ohio valley. Had this storm been an offshore noreaster type event, it might have been the biggest series of snow storms in the Boston area since 4 snowstorms of 1 foot or more hit in 9 days in late February to early March, 1707.

Sometimes extremes follow each other very quickly.
Chris - If the first Montpellier storm was a 100-year event, according to Meteo France as you reported, then the second storm's occurring within two days signifies a change in the underlying system behavior.

By definition, if the long-term weather cycle is holding, then such a storm will only occur once in a hundred years, as you know. When two such events occur in quick succession, the statisticians conclude that the underlying system's behavior has changed, because the likelihood of such behavior would be virtually zero* if it were holding. The statistics are incontrovertible; the only variable is whether Meteo France is correct about the first storm's designation.

So, as someone with no meteorology expertise but a number of years of statistical training and experience, I would conclude that if Meteo France is correct, then the weather in the Montpellier area of France is no longer following the 1960-2014 pattern, but has begun a new pattern.

Notes:

* (1/36,475) x (2/36,475) ~ 1 / 665,200,000. [There are 36,475 days in 100 years. The storms occurred 2 days apart.]

This is less than 1 chance of such a double-storm event in 500 million days, or once in 1,824,000 years. The Occam's Razor principle suggests that a more reasonable explanation is that the weather/climate cycle has changed, or that Meteo France's characterization of the storm was incorrect.

Oncor officials say storms that left 300,000 in the dark are among worst they've seen

Link

Wind speeds reached 90 mph.
Quoting 8. jaxter99:

Chris - If the first Montpellier storm was a 100-year event, according to Meteo France as you reported, then the second storm's occurring within two days signifies a change in the underlying system behavior.

By definition, if the long-term weather cycle is holding, then such a storm will only occur once in a hundred years, as you know. When two such events occur in quick succession, the statisticians conclude that the underlying system's behavior has changed, because the likelihood of such behavior would be virtually zero* if it were holding. The statistics are incontrovertible; the only variable is whether Meteo France is correct about the first storm's designation.

So, as someone with no meteorology expertise but a number of years of statistical training and experience, I would conclude that if Meteo France is correct, then the weather in the Montpellier area of France is no longer following the 1960-2014 pattern, but has begun a new pattern.

Notes:

* (1/36,475) x (2/36,475) ~ 1 / 665,200,000. [There are 36,475 days in 100 years. The storms occurred 2 days apart.]

This is less than 1 chance of such a double-storm event in 500 million days, or once in 1,824,000 years. The Occam's Razor principle suggests that a more reasonable explanation is that the weather/climate cycle has changed, or that Meteo France's characterization of the storm was incorrect.


Your calculation assumed that every day of the year is equally likely, which is not valid, because this is the time of year that such events are most likely to occur in the area. Also, note that the records weren't set at the same town or city both times. There are multiple towns in the area at which a similar event can occur, which increases the chances of this happening somewhere in the area, assuming that this is a 1-in-100 year event for Montpellier, and also for each of many of the other towns in the area. Finally, your first calculation should have probably been for the probability of the first event happening some time this year, instead of on that exact day, and then following with the probability that the next record was set so soon afterwards. In any event, I believe that the records were set about 13 days apart.

PS If I were doing the estimate, I would take the first record event as a given, because a similar record is likely to be set somewhere in the world, and then calculate the probability of the 2nd record event being so soon afterwards, given that the first event happened. So after the 1st event happened, I would put the probability of the 2nd event being in the same year (this is the time of year that such events are more likely there), at about 1 in 50 (since the record before this year was set about 35 years ago, but records get harder to break as new ones are set). It's similar with the probability of the same person winning the lottery twice in a row (I don't play lotteries). Take the first win as a given, because someone is likely to win. Then the probability of the same person winning on the next draw is the same as any given person who bought a ticket winning (I'm assuming here that each player buys exactly one ticket), so 1 in a few million (as opposed to 1 in a few million squared). So my ballpark estimate for the two rainfall records being in the same year is about 1 in 50. NOTE: I'm not implying that the probability of such events happening isn't changing.
Quoting 10. DCSwithunderscores:



Your calculation assumed that every day of the year is equally likely, which is not valid, because this is the time of year that such events are most likely to occur in the area. Also, note that the records weren't set at the same town or city both times. There are multiple towns in the area at which a similar event can occur, which increases the chances of this happening somewhere in the area, assuming that this is a 1-in-100 year event for Montpellier, and also for each of many of the other towns in the area. Finally, your first calculation should have probably been for the probability of the first event happening some time this year, instead of on that exact day, and then following with the probability that the next record was set so soon afterwards. In any event, I believe that the records were set about 13 days apart.

PS If I were doing the estimate, I would take the first record event as a given, because a similar record is likely to be set somewhere in the world, and then calculate the probability of the 2nd record event being so soon afterwards, given that the first event happened. So after the 1st event happened, I would put the probability of the 2nd event being in the same year (this is the time of year that such events are more likely there), at about 1 in 35 (since the record before this year was set about 35 years ago. It's similar with the probability of the same person winning the lottery twice in a row (I don't play lotteries). Take the first win as a given, because someone is likely to win. Then the probability of the same person winning on the next draw is the same as any given person who bought a ticket winning (I'm assuming here that each player buys exactly one ticket), so 1 in a few million (as opposed to 1 in a few million squared). So my ballpark estimate for the two rainfall records being in the same year is about 1 in 35. NOTE: I'm not implying that the probability of such events happening isn't changing.


yes i agree. Those assumptions are full of faults.
1-THIS IS THE TIME OF THE YEAR WHICH ALMOST ALWAYS WITNESSED RECORD RAINFALLS IN SOUTHERN FRANCE.
2-This year also witnessed a good combiantion with favorable meridian movements and warm sea temperatures.
3-The two events were held IN DIFFERENT PLACES. The fact the AREA it is the same it is because THAT area is prone to floods because its topography.
4-A particular rare combination of those events in a favorable year in the most favorable time of the year caused two similar events, so it is absolutely wrong to skip the climatological and meteorological factors and just use the nude statistics and assume 100x100=1 in 10000 years event, Not true. The same favorable conditions were in place during the past weeks.

Having said so, having both storms in the same year it was due to the same coincidence of favorable events and those events occurred in places prone to floods. In this case, having two IN DIFFERENT PLACES instead one doesn't change much the time of return.

Given all favorable conditions in place, there might be 1 chance every 3-4 to have two events like this in the same year. So the real time of return would be 3-4 centuries not millions of years.

But I still think it is too much, since sea temperature is higher now than decades or centuries ago , the average atmospheric temperature is higher,so the energy at stake is higher, so assuming a hypotetical stability of the climate in the medium-long term (which doesn t occurr) I think the return period for a double event of this magnitude in different places of that area is no more than 100 years,probably less.

Still a remarkable event, but not something like millions or even thousands of years event.
That would be if the storm were at the same place, but they were NOT. They were at different places in THE EXACT AREA which is prone to floods.


weatherhistorian has created a new entry.
Quoting 10. DCSwithunderscores

Your calculation assumed that every day of the year is equally likely, which is not valid, because this is the time of year that such events are most likely to occur in the area.


Thanks for the correction about the probability for particular days in a year - I agree the probability should be based on the seasonal likelihood. As Chris pointed out, this kind of rainfall typically occurs in September and October, amounting to a period of about 60 days. I use this (rather than 364.75) as an annual average # of eligible days for the updated calculations.

Quoting 10. DCSwithunderscores

Finally, your first calculation should have probably been for the probability of the first event happening some time this year, instead of on that exact day, and then following with the probability that the next record was set so soon afterwards. In any event, I believe that the records were set about 13 days apart.


The probability of the Montpellier event occurring at any time in the 60 day period in 2014 is 1 in 100, for the entire period - this reflects the definition of a 100-year storm. So, I will amend the calculation to reflect the likelihood of the Montpellier event's being 1:100 for 2014.

Quoting 10. DCSwithunderscores

Also, note that the records weren't set at the same town or city both times. There are multiple towns in the area at which a similar event can occur, which increases the chances of this happening somewhere in the area, assuming that this is a 1-in-100 year event for Montpellier, and also for each of many of the other towns in the area.


The definition of a 100-year flood is usually set for the entire floodplain. Looking at a topographic map for the southeast of France, it appears that although Saint-Gervais-sur-Mare is in the foothills, it is in the same floodplain area as Montpellier, not a different area. They are only 50 miles apart. So, the two would almost certainly have coupled probability distributions for storm predictions, not independent ones.

Quoting 10. DCSwithunderscores

In any event, I believe that the records were set about 13 days apart.


Absolutely correct - my error. But the calculation should then reflect all of the likelihoods that the 2 storms could have occurred on any of the 13 days in the interval, not just the 13th. I.e. It should use the cumulative probability for all of these days.

Quoting 10. DCSwithunderscores

[...calculation details omitted...jaxter99]
So my ballpark estimate for the two rainfall records being in the same year is about 1 in 50.


Ummm, wait a minute - you're saying that the probability of 2 such events occurring in the same year (1 in 50) is more likely than the probability of a single such event (1 in 100)?? Intuitively, that can't possibly be right.

I think the error is caused by assuming the first event has happened, and that it was a certainty. If we were only concerned with the probability of the second event, you would be right. But we're trying to determine the probability of both events together. In that case, the first event was not a certainty; by definition, it had an annual likelihood of 1 in 100. The second is the same - 1 in 100 for the same year.

But that's not sufficient. These storms occurred 13 days apart, not just in the same year, so the shortness of that interval should be factored in.

The revised calculations* yield the result of 1 in 46,000, or 0.0022%. This is not impossible, but it's way outside the 3 standard deviations confidence interval a statistician would apply to determine whether the sample data fit the assumed distribution (1 such storm in 100 years). This signifies that, with better than 99% certainty, the underlying system is no longer behaving the same.

So, the conclusion must be the same: Either the weather/climate cycle has changed, or Meteo France made an error in characterizing the storm as a 100-year storm.

Notes:
* The likelihood that the first storm occurred on any day in the 2014 season is 1 in 100 (approximately), per Meteo France.
The probability that another, similar storm would occur within the next 13 days is the cumulative probability for each of the days in the interval. This is (approximately) 13/60 x 1/100, assuming there are 60 days in the season, and the probability of a second such storm is still 1/100.
So, pr(2 storms, 13 days apart) = (1/100) x (13/60) x (1/100) ~ 1/46,000.

This is based on the assumption that the Saint-Germain-sur-Mare event was also a 100-year event, because it set a record. That may be pessimistic, since records have only been kept for the past 50 years, and a similar (or worse) storm might have occurred between 1910 and 1960. However, the net effect if the assumption is wrong is virtually the same; the revised probability would be (at its most likely) 1/23,000, or 0.0044%. This is still well outside the 99% confidence interval, leading to the same conclusion.
Quoting 14. jaxter99:



Thanks for the correction about the probability for particular days in a year - I agree the probability should be based on the seasonal likelihood. As Chris pointed out, this kind of rainfall typically occurs in September and October, amounting to a period of about 60 days. I use this (rather than 364.75) as an annual average # of eligible days for the updated calculations.



The probability of the Montpellier event occurring at any time in the 60 day period in 2014 is 1 in 100, for the entire period - this reflects the definition of a 100-year storm. So, I will amend the calculation to reflect the likelihood of the Montpellier event's being 1:100 for 2014.



The definition of a 100-year flood is usually set for the entire floodplain. Looking at a topographic map for the southeast of France, it appears that although Saint-Gervais-sur-Mare is in the foothills, it is in the same floodplain area as Montpellier, not a different area. They are only 50 miles apart. So, the two would almost certainly have coupled probability distributions for storm predictions, not independent ones.



Absolutely correct - my error. But the calculation should then reflect all of the likelihoods that the 2 storms could have occurred on any of the 13 days in the interval, not just the 13th. I.e. It should use the cumulative probability for all of these days.



Ummm, wait a minute - you're saying that the probability of 2 such events occurring in the same year (1 in 50) is more likely than the probability of a single such event (1 in 100)?? Intuitively, that can't possibly be right.

I think the error is caused by assuming the first event has happened, and that it was a certainty. If we were only concerned with the probability of the second event, you would be right. But we're trying to determine the probability of both events together. In that case, the first event was not a certainty; by definition, it had an annual likelihood of 1 in 100. The second is the same - 1 in 100 for the same year.

But that's not sufficient. These storms occurred 13 days apart, not just in the same year, so the shortness of that interval should be factored in.

The revised calculations* yield the result of 1 in 46,000, or 0.0022%. This is not impossible, but it's way outside the 3 standard deviations confidence interval a statistician would apply to determine whether the sample data fit the assumed distribution (1 such storm in 100 years). This signifies that, with better than 99% certainty, the underlying system is no longer behaving the same.

So, the conclusion must be the same: Either the weather/climate cycle has changed, or Meteo France made an error in characterizing the storm as a 100-year storm.

Notes:
* The likelihood that the first storm occurred on any day in the 2014 season is 1 in 100 (approximately), per Meteo France.
The probability that another, similar storm would occur within the next 13 days is the cumulative probability for each of the days in the interval. This is (approximately) 13/60 x 1/100, assuming there are 60 days in the season, and the probability of a second such storm is still 1/100.
So, pr(2 storms, 13 days apart) = (1/100) x (13/60) x (1/100) ~ 1/46,000.

This is based on the assumption that the Saint-Germain-sur-Mare event was also a 100-year event, because it set a record. That may be pessimistic, since records have only been kept for the past 50 years, and a similar (or worse) storm might have occurred between 1910 and 1960. However, the net effect if the assumption is wrong is virtually the same; the revised probability would be (at its most likely) 1/23,000, or 0.0044%. This is still well outside the 99% confidence interval, leading to the same conclusion.


For the purpose of determining whether or not the climate is changing, I would calculate the probability of both events being in the same year, but not of both being this year, as I think that the year that it happens is less important than that two are together. I prefer that way. When I did that estimate, I was interested more in the national record than that region in particular, so I used the duration since the national record was last broken, along with an "adjustment" for the record becoming harder to beat after the Saint-Gervais-Sur-Marie event.

I wouldn't take into account that they were only 13 days apart, because if they are in the same year then of course they are likely to be during the time that these kind of events are most likely - in other words, these events don't follow a uniform distribution throughout the year, so I wouldn't assume one.

To me the 1-in-100 year event status is ambiguous, because one needs a context of over what specific region, or is it at just one station? I took it as being at that one station. Regardless, I thought that the duration since the 1979 national record was more reliable for my estimate.

If I were to calculate the probability of the 2 top events being both this year (which you prefer), instead of just being in the same year, I would take a 50 year period (assuming that 2-hour precipitation records have been kept for about that long), then the probability of both being this year would be about (1/50)^2, or about 1 in 2500, but this ignores the possibility that conditions this year are more favorable for these events (independent of climate change), which would increase the probability of two of them coming in the same year.

Regardless, I don't consider calculating the probability of both being this year (instead of being just in the same year), useful for climate change purposes, because there are going to be records set anyways with so many places and types of records to choose from, so it is not at all remarkable that the Saint-Gervais-Sur-Mare record occurred this year. To me the only significantly remarkable fact (for this purpose) is that there was another in the same year. So I'm going to stick with my (crude) estimate of 1 in 50, for the purpose at hand.

You can calculate your estimate as you please.
Jaxter99: PS 50 people throw a number each into a hat for a draw. A number is drawn and person X wins. This is not at all remarkable, because someone is bound to win. Another draw with the same people is held again the following week. If person X wins again, then that is somewhat remarkable (but this kind of thing can happen). To me, it wouldn't make much sense to calculate the probability that person X wins both times, because someone is bound to win the first time. If someone asked in advance, before the first draw, what is the probability of person X winning both times, then one would use both times in the calculation, but after the fact, it isn't significant that person X won the first time. Similarly, national records are bound to happen every year. France happened to "win" with a 2-hour precipitation record. That isn't remarkable after the fact, but if France "wins" again the same year, then to me it makes sense to calculate the probability of that happening again, given that it already happened this year. That's the way that I see it. It's not the only valid way, as long as readers understand what is being calculated and why. Readers can decide which they prefer, your method or mine (if any are still reading here).
And again.
https://www.facebook.com/KeraunosObs
Quoting 17. cRRKampen:

And again.
https://www.facebook.com/KeraunosObs


yes, again: 270mm in Prades le Lez yesterday in a few hours, Montpellier flooded again.
Quoting 18. mottoth:



yes, again: 270mm in Prades le Lez yesterday in a few hours, Montpellier flooded again.

Sounds a bit overly alarmistic... It was 269.9mm ;)
And again, number four: http://www.lepetitjournal.com/international/france -monde/actualite/196505-intemperies-meteo-france-p lace-le-gard-en-vigilance-rouge
21. vis0
CREDIT:: NOAA, presented via aviationweather.org
SUBJECT::Big LOW W of Europe dragging 1st generated remnants of Fay towards France???

+++having trouble utube not providing embeds (for my science Vids),. here the link http://youtu.be/frql9f_ru3E