dynamicists versus modellers: a growing divide?

2
94 Weather – April 2010, Vol. 65, No. 4 Heavy rainfall in long time series throughout the Ryedale record, and for more heavy rain days to occur in summer than in other seasons. When compared with the most recent period (2001–2008), higher fre- quencies of heavy rain days and a greater proportion of annual rainfall falling on heavy rain days occurred in the 1930s. The composite record which has been con- structed for Ryedale in this study has enabled heavy rainfalls and seasonal trends to be assessed over a 93-year period in an upland area. The conclusion that recent heavy rain- fall events, particularly in summer, are not as rare as may first have been thought clearly illustrates the importance of using long-term climatic monitoring and associated records to put recent weather events into context. Acknowledgements We gratefully acknowledge use of the British Atmospheric Data Centre and also the use of a synoptic chart (both UK Met Office), and the use of the Digimap service and Strategi® map data. Jeff Warburton also acknowledges support from NERC (Award NE/D005744/1). References Aguilar E, Auer I, Brunet M, Peterson TC, Wieringa J. 2003. Guidelines on climate metadata and homogenization. World Climate Data and Monitoring Programme Report 53 (WMO/TD 1186). World Meteorological Organisation: Geneva. Aron G, Rachford TM. 1974. Procedures for filling gaps in hydrological event series. Water Resources Bull 10: 719–727. Burt S. 2005. Cloudburst upon Hendraburnick Down: The Boscastle storm of 16 August 2004. Weather 60: 219–227. Burt TP. 2009. Homogenising the rainfall record at Durham for the 1870s. Hydrolog. Sci. J. 54: 199–209. Burt TP, Horton BP. 2007. Inter-decadal variability in daily rainfall at Durham (UK) since the 1850s. Int. J. Climatol. 27: 945–956. Centre for Ecology and Hydrology, Wallingford. 2005. 27055 – Rye at Broadway Foot Rainfall. Available at http:// www.nerc-wallingford.ac.uk/ih/nrfa/ spatialinfo/Rainfall/rainfall027055.html [Accessed 6 April 2009]. Cinderey M. 2003. The North Yorkshire- Teesside storm of 10 August 2003. Weather 60: 60–65. Cinderey M. 2005. North York Moors storms – 19 June 2005. Weather 60: 273. Fowler HJ, Kilsby CG. 2003. A regional frequency analysis of United Kingdom extreme rainfall from 1961–2000. Int. J. Climatol. 23:1313–1334. Gruntfest E, Handmer J. 2001. Dealing with flash floods: contemporary issues and future possibilities, in Coping with flash floods. Gruntfest E, Handmer J (eds). Kluwer Academic Publishers: Dordrecht. pp 3–10. Jenkins GJ, Perry MC, Prior MJO. 2008. The climate of the United Kingdom and recent trends. Revised edition, January 2009. Met Office Hadley Centre: Exeter. Karl TR, Knight RW. 1998. Secular trends of precipitation amount, frequency and intensity in the United States. Bull. Am. Meteorol. Soc. 79: 231–241. Lane SN. 2008. Climate change and the summer 2007 floods in the UK. Geography 93: 91–97. Malby AR, Whyatt JD, Tummis RJ, Orr HG. 2007. Long-term variations in orographic rainfall: analysis and implica- tions for upland catchments. Hydrol. Sci. J. 52: 276–291. Maraun D, Osborn TJ, Gillett NP. 2008. United Kingdom daily precipitation inten- sity: improved early data, error estimates and an update from 2000 to 2006. Int. J. Climatol. 28: 833–842. Marsh TJ, Hannaford J. 2007. The sum- mer 2007 floods in England and Wales – a hydrological appraisal. Centre for Ecology and Hydrology: Wallingford. Correspondence to: Jonathan Hopkins, Durham University, Department of Geography, Science Laboratories, South Road, Durham, DH1 3LE, UK. [email protected] © Royal Meteorological Society, UK DOI: 10.1002/wea.552 Murphy JM, Sexton DMH, Jenkins GJ, Boorman PM, Booth BBB, Brown CC, Clark RT, Collins M, Harris GR, Kendon EJ, Betts RA, Brown SJ, Howard TP, Humphrey KA, McCarthy MP, McDonald RE, Stephens A, Wallace C, Warren R, Wilby R, Wood RA. 2009. UK Climate Projects Science Report: Climate change pro- jections. Met Office Hadley Centre: Exeter. Osborn TJ, Hulme M. 2002. Evidence for trends in heavy rainfall events over the UK. Philos. T. Roy. Soc. A 360: 1313–1325. Osborn TJ, Hulme M, Jones PD, Basnett TA. 2000. Observed trends in the daily intensity of United Kingdom precipitation. Int. J. Climatol. 20: 347–364. Rodda HJE, Little MA, Wood RG, MacDougall N, McSharry PE. 2009. A digital archive of extreme rainfalls in the British Isles from 1866 to 1968 based on British Rainfall. Weather 64: 71–75. Sibley AM. 2009. Analysis of the North York Moors storms – 19 June 2005. Weather 64: 39–42. Walker JP. 2008. The discontinuance of Boltby reservoir, North Yorkshire, UK. Dams and Reservoirs 18: 17–21. Wass P, Lindsay D, Faulkner D. 2008. Flash Flood! A lucky escape for 10,000 bikers. Paper presented to British Hydrological Society 10th National Hydrology Symposium, Exeter, 2008. UK Met Office. 2006. MIDAS Land Surface Stations data (1853–current). British Atmospheric Data Centre 2006. Available at http://badc.nerc.ac.uk/data/ukmo-midas [Accessed 7th November 2009]. Dynamicists versus modellers: a growing divide? Meeting report This meeting, on 18 November 2009, was organised by the Society’s History Group and was held at the University of Reading. It was advertised with what was described during the meeting as a deliberately pro- vocative abstract, asking whether the mete- orological community has forgotten the pioneering principles of Rossby and Charney and whether today’s modellers are really computer scientists who ‘tweak’ models without fully understanding them. In his introduction, Malcolm Walker (Chair of the History Group) introduced the cast of characters (Rossby, Charney, Sutcliffe et al.) whose classic papers in the 1940s laid the foundations of dynamical meteorology, and whose experiments with numerical weather prediction (NWP) led to the first computer- based weather forecasting systems. The sec- ond speaker, John Methven (University of Reading), gave an introduction to the early theories of the development of extratropical weather systems, beginning with the Bergen School and the development of the polar front theory of cyclones and ending with the independent development of quasi- geostrophic theory by Rossby and Charney. The point was stressed that what both Rossby and Charney sought, and indeed found, was a mathematically closed theory for the evolution of synoptic-scale cyclones. The initial NWP systems were based on quasi-geostrophic equations, keeping a close link with theory. With increases in computing power, and the realisation that accurate prediction at finer resolution required greater knowledge of the unbal- anced flow, NWP moved to the primitive equations and the link to theory began to weaken. The talk ended with a Monty Python inspired question: what has quasi- geostrophic theory ever done for us? John’s answer: It has given us the ability to under- stand the evolving flow and develop intel- ligent diagnostics for analysing the output from numerical simulations. Next, Lennart Bengtsson (University of Reading) gave a summary of the initial development of NWP. His first point fol- lowed directly from John Methven’s talk: the fundamental requirement for the inception

Upload: thomas-h-a-frame

Post on 06-Jul-2016

214 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Dynamicists versus modellers: a growing divide?

94

Wea

ther

– A

pril

2010

, Vol

. 65,

No.

4H

eavy

rain

fall

in lo

ng t

ime

seri

es

throughout the Ryedale record, and for more

heavy rain days to occur in summer than in

other seasons. When compared with the

most recent period (2001–2008), higher fre-

quencies of heavy rain days and a greater

proportion of annual rainfall falling on heavy

rain days occurred in the 1930s.

The composite record which has been con-

structed for Ryedale in this study has enabled

heavy rainfalls and seasonal trends to be

assessed over a 93-year period in an upland

area. The conclusion that recent heavy rain-

fall events, particularly in summer, are not as

rare as may first have been thought clearly

illustrates the importance of using long-term

climatic monitoring and associated records

to put recent weather events into context.

Acknowledgements

We gratefully acknowledge use of the British

Atmospheric Data Centre and also the use of

a synoptic chart (both UK Met Office), and

the use of the Digimap service and Strategi®

map data. Jeff Warburton also acknowledges

support from NERC (Award NE/D005744/1).

ReferencesAguilar E, Auer I, Brunet M, Peterson TC, Wieringa J. 2003. Guidelines on climate metadata and homogenization. World Climate Data and Monitoring Programme Report 53 (WMO/TD 1186). World Meteorological Organisation: Geneva.

Aron G, Rachford TM. 1974. Procedures for filling gaps in hydrological event series. Water Resources Bull 10: 719–727.

Burt S. 2005. Cloudburst upon Hendraburnick Down: The Boscastle storm of 16 August 2004. Weather 60: 219–227.

Burt TP. 2009. Homogenising the rainfall record at Durham for the 1870s. Hydrolog. Sci. J. 54: 199–209.

Burt TP, Horton BP. 2007. Inter-decadal variability in daily rainfall at Durham (UK) since the 1850s. Int. J. Climatol. 27: 945–956.

Centre for Ecology and Hydrology, Wallingford. 2005. 27055 – Rye at Broadway Foot Rainfall. Available at http://www.nerc-wallingford.ac.uk/ih/nrfa/spatialinfo/Rainfall/rainfall027055.html [Accessed 6 April 2009].

Cinderey M. 2003. The North Yorkshire-Teesside storm of 10 August 2003. Weather 60: 60–65.

Cinderey M. 2005. North York Moors storms – 19 June 2005. Weather 60: 273.

Fowler HJ, Kilsby CG. 2003. A regional frequency analysis of United Kingdom extreme rainfall from 1961–2000. Int. J. Climatol. 23:1313–1334.

Gruntfest E, Handmer J. 2001. Dealing with flash floods: contemporary issues and future possibilities, in Coping with flash floods. Gruntfest E, Handmer J (eds). Kluwer Academic Publishers: Dordrecht. pp 3–10.

Jenkins GJ, Perry MC, Prior MJO. 2008. The climate of the United Kingdom and recent trends. Revised edition, January 2009. Met Office Hadley Centre: Exeter.

Karl TR, Knight RW. 1998. Secular trends of precipitation amount, frequency and intensity in the United States. Bull. Am. Meteorol. Soc. 79: 231–241.

Lane SN. 2008. Climate change and the summer 2007 floods in the UK. Geography 93: 91–97.

Malby AR, Whyatt JD, Tummis RJ, Orr HG. 2007. Long-term variations in orographic rainfall: analysis and implica-tions for upland catchments. Hydrol. Sci. J. 52: 276–291.

Maraun D, Osborn TJ, Gillett NP. 2008. United Kingdom daily precipitation inten-sity: improved early data, error estimates and an update from 2000 to 2006. Int. J. Climatol. 28: 833–842.

Marsh TJ, Hannaford J. 2007. The sum-mer 2007 floods in England and Wales – a hydrological appraisal. Centre for Ecology and Hydrology: Wallingford.

Correspondence to: Jonathan Hopkins,Durham University, Department of Geography, Science Laboratories, South Road, Durham, DH1 3LE, UK.

[email protected]

© Royal Meteorological Society, UK

DOI: 10.1002/wea.552

Murphy JM, Sexton DMH, Jenkins GJ, Boorman PM, Booth BBB, Brown CC, Clark RT, Collins M, Harris GR, Kendon EJ, Betts RA, Brown SJ, Howard TP, Humphrey KA, McCarthy MP, McDonald RE, Stephens A, Wallace C, Warren R, Wilby R, Wood RA. 2009. UK Climate Projects Science Report: Climate change pro-jections. Met Office Hadley Centre: Exeter.

Osborn TJ, Hulme M. 2002. Evidence for trends in heavy rainfall events over the UK. Philos. T. Roy. Soc. A 360: 1313–1325.

Osborn TJ, Hulme M, Jones PD, Basnett TA. 2000. Observed trends in the daily intensity of United Kingdom precipitation. Int. J. Climatol. 20: 347–364.

Rodda HJE, Little MA, Wood RG, MacDougall N, McSharry PE. 2009. A digital archive of extreme rainfalls in the British Isles from 1866 to 1968 based on British Rainfall. Weather 64: 71–75.

Sibley AM. 2009. Analysis of the North York Moors storms – 19 June 2005. Weather 64: 39–42.

Walker JP. 2008. The discontinuance of Boltby reservoir, North Yorkshire, UK. Dams and Reservoirs 18: 17–21.

Wass P, Lindsay D, Faulkner D. 2008. Flash Flood! A lucky escape for 10,000 bikers. Paper presented to British Hydrological Society 10th National Hydrology Symposium, Exeter, 2008.

UK Met Office. 2006. MIDAS Land Surface Stations data (1853–current). British Atmospheric Data Centre 2006. Available at http://badc.nerc.ac.uk/data/ukmo-midas [Accessed 7th November 2009].

Dynamicists versus modellers: a growing divide?Meeting reportThis meeting, on 18 November 2009, was

organised by the Society’s History Group

and was held at the University of Reading.

It was advertised with what was described

during the meeting as a deliberately pro-

vocative abstract, asking whether the mete-

orological community has forgotten the

pioneering principles of Rossby and Charney

and whether today’s modellers are really

computer scientists who ‘tweak’ models

without fully understanding them.

In his introduction, Malcolm Walker (Chair

of the History Group) introduced the cast of

characters (Rossby, Charney, Sutcliffe et al.)

whose classic papers in the 1940s laid the

foundations of dynamical meteorology, and

whose experiments with numerical weather

prediction (NWP) led to the first computer-

based weather forecasting systems. The sec-

ond speaker, John Methven (University of

Reading), gave an introduction to the early

theories of the development of extratropical

weather systems, beginning with the Bergen

School and the development of the polar

front theory of cyclones and ending with

the independent development of quasi-

geostrophic theory by Rossby and Charney.

The point was stressed that what both

Rossby and Charney sought, and indeed

found, was a mathematically closed theory

for the evolution of synoptic-scale cyclones.

The initial NWP systems were based on

quasi-geostrophic equations, keeping a

close link with theory. With increases in

computing power, and the realisation that

accurate prediction at finer resolution

required greater knowledge of the unbal-

anced flow, NWP moved to the primitive

equations and the link to theory began to

weaken. The talk ended with a Monty

Python inspired question: what has quasi-

geostrophic theory ever done for us? John’s

answer: It has given us the ability to under-

stand the evolving flow and develop intel-

ligent diagnostics for analysing the output

from numerical simulations.

Next, Lennart Bengtsson (University of

Reading) gave a summary of the initial

development of NWP. His first point fol-

lowed directly from John Methven’s talk: the

fundamental requirement for the inception

Page 2: Dynamicists versus modellers: a growing divide?

95

Weather –

April 2010, Vol. 65, No. 4

of NWP was the theoretical basis provided

by Charney and Rossby. He noted that the

first operational NWP forecasts began in

December 1954; they were for the North

Atlantic region and were based on the baro-

tropic model developed by the Stockholm

group led by Rossby. The BESK computer

used for these initial forecasts had only a

512-word memory and could perform a sin-

gle operation approximately every 50

microseconds. Lennart pointed out that the

huge improvements in computer technol-

ogy since those first operational forecasts

have led to models becoming so complex

that it is difficult for users to know how to

interact with them. The next speaker, Brian

Golding (Met Office), sought in his presenta-

tion to answer the question: NWP since the

1960s: A triumph of numerical analysis or

meteorological science? He drew a distinc-

tion between NWP and meteorology: the

former is based on the numerical solution

of known physical laws and the latter is the

study of emergent weather phenomena. He

noted that there have been great improve-

ments in the skill of NWP, due to improve-

ments in the numerical models and the use

of observations to initialise the forecasts,

with current four-day forecasts as accurate

as one-day forecasts were around 1975. But

he stressed that these improvements could

only be judged and guided by comparison

with the observed emergent weather

phenomena.

The final speaker, Andy White (Met Office),

began with his favourite equation (the zonal

momentum equation on the sphere), stat-

ing that there are two options when faced

with it: you either immediately proceed to

a numerical solution or you carefully analyse

the equation discarding small terms and

reducing it to a simpler, more easily under-

stood, form. Both options have merits and

drawbacks. In the first, you obtain an accu-

rate solution but little understanding, whilst

in the second, understanding comes at the

expense of accuracy. He felt there was a

creative tension between those who prefer

each option, which is beneficial to both.

A panel and audience discussion followed

and was chaired by Brian Hoskins (University

of Reading/Grantham Institute for Climate

Change), and a lively debate centred on the

question: Is there a growing divide? The

large variety of opinions expressed indi-

cated that scientists in general are always

ready to talk to each other, even if they

cannot reach a consensus. On the one hand,

many were of the opinion that the sheer

size and complexity of modern computers

and models is such that there is inevitably

a larger number of computer technicians

and programmers who may have only a

limited understanding of dynamical mete-

orology. Some felt that modellers should be

dynamicists at heart, and voiced their con-

cern over the fact that new generations of

scientists increasingly use models without

developing an understanding of the under-

lying dynamics. One member of the panel

felt that some dynamical concepts, such as

group velocity, have simply been forgotten

and one audience member went so far as

to state that dynamicists were a dying breed.

Others were not so pessimistic, with some

feeling that NWP models have become

important tools which are used by dynami-

cists for knowledge generation. Some felt

that the divide was greater in the early days

of NWP, in part because of a divide that

existed between dynamicists and forecast-

ers. Several people raised the issue that the

increased emphasis on climate and statisti-

cal analysis has led to a reduction in the

amount of dynamical meteorology research

being performed, so that the real divide was

between dynamicists/modellers and statis-

ticians. Some felt that the sheer complexity

of the interaction between the dynamics

and physics included in modern NWP mod-

els means that no single person is capable

of understanding them as a whole, and that

the future of dynamical meteorology lies in

guiding our understanding of the increas-

ingly complex model world by the develop-

ment of intelligent diagnostics.

Whatever its future within NWP, it was

clear from the discussion that all agreed

that an understanding of dynamical mete-

orology should remain a key objective in

the education of young atmospheric

scientists.

Correspondence to: Thomas H. A. Frame, Meteorology Department,University of Reading,PO Box 243, Earley Gate, Reading RG6 6BB, UK.

[email protected]

© Royal Meteorological Society, 2010

DOI: 10.1002/wea.567

The long road aheadMeeting reportThis is a report of the joint RMetS and Royal

Society meeting on The Science of UK Climate

Projection: UKCP09 and Beyond, held on 15

October 2009 at the Royal Society.

While international political negotiations

seek to limit emissions of greenhouse gases,

climate change is already happening and

some level of future climate change will be

inevitable. How do we plan for the future?

This meeting discussed how to provide

information about regional climate change

on scales relevant for local and national

responses.

Because we require information at local

scales, and because we are interested in how

climate change will affect phenomena such

as rainfall, storms, heatwaves, etc., it is natural

to turn to complex climate models which are

designed to simulate the interactions

between the different processes that deter-

mine local climate. Despite the extensive use

of climate models in research and in putting

the scientific case for anthropogenic climate

change so far, imperfections in models mean

that projections are uncertain; different mod-

els produce different futures.

Recognising that we are not in possession

of a perfect model of the climate system,

and because there are natural fluctuations

in climate that cannot be deterministically

predicted, the most recent set of UK Climate

Projections, UKCP09, is presented in the

form of probabilities (http://ukclimatepro-

jections.defra.gov.uk; Street et al., 2009). The

method on which UKCP09 is based involves

exploring the input parameter space of a

model, in this case HadCM3, to find model

versions which simulate historical climates

consistent with available observations of

the climate system and their uncertainties.

The input parameters are those which con-

trol many of the physical processes and

feedbacks in the model, together with forc-

ing parameters, such as increasing green-

house gases. The model versions which best

simulate historical climate are used to form

probability distribution functions (PDFs) of

future climate change.

Despite this relatively simple concept,

implementation of the UKCP09 projections

is very complex. Limitations on computer

resources mean that the projections are

constructed from a fairly limited number of

samples of the model input parameter

space – even though those 400 or so

ensemble simulations took a team of Hadley

Centre scientists over eight years to pro-

duce. Considerable effort is required to turn

this ensemble into a more comprehensive

sample of model input parameter space

that can be compared with the available

observations, including fitting statistical

models called emulators to mimic the out-

put of HadCM3. In many cases, quantitative

estimates of uncertainty or errors in obser-

vations are not available, so assessing the

extent to which model versions actually

compare to the real world is a challenge.

Some assumptions have to be made in the

implementation of the UKCP09 algorithm

for reasons of expediency; nevertheless,

the sensitivity of the final PDFs to many

Meeting report