Ensemble forecasts

[franksingleton]

From time to time, I post without comment, Met `Office Blog items that might be of more than passing interest. However, Understanding ensemble forecasting: How the Met Office predicts uncertainty Is, I think, rather better and more interesting than the general run.

 

[LittleSister]

I found that interesting and informative, and answered questions I already had in mind after often seeing various outputs from ensemble models (on e.g. the Met Office's '10 Day Trend' videos)..

A pedant writes:
While it is generally very commendably clearly and carefully written, I was struck by one sentence that stood out to me as different, seeming to suggest an inordinate correspondence between the model and reality:
'The higher resolution of the UK ensemble allows for more detailed predictions, while the global ensemble ensures that large-scale weather patterns are accurately represented.' [emphasis added]

This seems at odds with the uncertainties and limitations emphasised in the rest of the piece, for example:
'While ensemble forecasts provide valuable guidance, they cannot offer a perfect representation of probability.'

Am I reading that first quoted sentence wrongly?

 

[Stemar]

My reading of it is that the global ensemble give a big picture of what's happening in the atmosphere, while the UK one digs into more detail, using the global as a starting point.

AIUI, big weather patterns can be fairly accurately predicted - a depression in the Atlantic, driven by the jet stream, is going to mean wind and rain in the UK, but a 70km grid is never going to tell us if the rain will fall on Bristol or Birmingham a week ahead.

 

[franksingleton]

I found that interesting and informative, and answered questions I already had in mind after often seeing various outputs from ensemble models (on e.g. the Met Office's '10 Day Trend' videos)..

A pedant writes:
While it is generally very commendably clearly and carefully written, I was struck by one sentence that stood out to me as different, seeming to suggest an inordinate correspondence between the model and reality:
'The higher resolution of the UK ensemble allows for more detailed predictions, while the global ensemble ensures that large-scale weather patterns are accurately represented.' [emphasis added]

This seems at odds with the uncertainties and limitations emphasised in the rest of the piece, for example:
'While ensemble forecasts provide valuable guidance, they cannot offer a perfect representation of probability.'

Am I reading that first quoted sentence wrongly?

I am always wary when the word accurate is used in meteorology. I would have preferred “….. well predicted …”

I can see the value of detailed ensembles for assessment of risk in areas particularly prone to flash flood, such as Lynmouth and Boscadtle.

 

[franksingleton]

My reading of it is that the global ensemble give a big picture of what's happening in the atmosphere, while the UK one digs into more detail, using the global as a starting point.

AIUI, big weather patterns can be fairly accurately predicted - a depression in the Atlantic, driven by the jet stream, is going to mean wind and rain in the UK, but a 70km grid is never going to tell us if the rain will fall on Bristol or Birmingham a week ahead.

Agree absolutely. To me, having joined the Met Office 70 years ago, the ability to predict severe storms several days ahead represent a remarkable improvement. When it comes to detail, precise track and development, there will always be uncertainty. Whether using physics based models or AI, a major limitation will remain detail below the resolution of observing systems and the chaos effect.

 

[LittleSister]

I can see the value of detailed ensembles for assessment of risk in areas particularly prone to flash flood, such as Lynmouth and Boscadtle.

One would have to look beyond weather prediction to best forecast flooding in such places.

There are lots of other factors involved, not least the building of houses/shops in what is more or less the river bed! (The flood plain is highly restricted in width at Lynmouth and similar places, so in periods of high flow the 'excess' water can largely only increase in depth, at the very same time the flow is reaching highly destructive speeds.

I've forgotten all the detail now, but I had a peripheral professional involvement with the Lynmouth situation decades ago. IIRC the 1952 flood disaster was the result of not just very unusually intense rainfall on that particular part of Exmoor that drains to Lynemouth, but also that the moor was already waterlogged, trees and branches were swept down the river, some of these became jammed against bridges partially blocking the flow, floodwater built up behind these, and when the bridge(s) gave way that released a sudden huge pulse of floodwater that caused much of the death and destruction. I think there may also have been an unfortunate coincidence in the timing of the peak flows coming down the West and East Lynn Rivers arriving at their confluence around the same time.

Exmoor was once forested, and this would have substantially absorbed/slowed the rain's path to the streams and river. The forest was gradually cleared beginning in prehistoric times, the moors became boggy, and in 19th century drains were built in a (not very successful) attempt to make the boggy parts of the moors more productive for agriculture, and these drains speed the flow of water off the moor into the rivers. (There was an ongoing programme a few decades ago (which may be continuing) to remove/block some of these old drains, to reduce flood risk downstream and support biodiversity on the moor.)

The design of the river bridge(s) between the moor and the town had meant they were susceptible to blockage by trees and debris swept down the river, yet not strong enough to hold the resulting (albeit unusual) build up of water behind. (The loss of the bridge(s) hindered rescue efforts afterwards.) The design of the replacement bridge(s) sought to address those problems.

I imagine that there is now flood depth and flow monitoring upstream of the town and warning systems in place, as well as better and, as you mention, more targetted forecasting.

Again, I can't remember the details, but the weather at the time was very exceptional in the amount of rain that fell in a localised area(s), even taking into account the how wet Exmoor can be. (Something about an intense front passing over, then moving backwards over it again?)

Of course we never did, still don't, and can't build our development/infrastructure/houses/towns to be able to cope with all exceptional weather and other natural phenomena. The only questions being how exceptional could be coped with/tolerated, and the balance of risk and the opportunity costs of avoidance.

 

[franksingleton]

The Met Office may be the spokesperson for flood warnings but these are a joint effort. I am not sure about Claudia where there is a vast amount of rain and flooding is a nigh certainty but, otherwise, flood prediction is a joint effort. Rain and/or snow melt are the inputs but the effects are a matter for hydrologists. That is why the radar network is jointly funded by the Met Office and the Department for `science and Environment Agency.At one time it was on my budget.

But what you say is correct. Land use has made flooding mote likely.