• No results found

ISTI generic lo res

N/A
N/A
Protected

Academic year: 2020

Share "ISTI generic lo res"

Copied!
37
0
0

Loading.... (view fulltext now)

Full text

(1)

Ini$a$ve  overview  

08/19/11  

(2)

Talk  Outline  

Background  

Data  rescue  and  databank  

Mul6plicity  of  data  products  

Common  assessment  through  consistent  

performance  benchmarking  

Data  and  product  portal  development  

(3)

Background  

Several  independent  global  es6mates  of  Land  

Surface  Air  Temperatures  

Independently  produced  

Gross  agreement  on  all  6mescales  

Process  metadata  in  some  cases  missing  

Much  of  the  work  was  undertaken  in  late  1980’s/  

early  1990’s  -­‐  technology  and  expecta6ons  have  

changed  since  then  

Monthly  large  scale  gridbox  averages  only  (generally)  

(4)

Examples  of  temperature  reconstruc$ons  

IPCC Fourth Assessment Report (2007), Figure 3.1

Annual anomalies of global land-surface air temperature (°C), 1850 to 2005, relative to the 1961 to 1990 mean for CRUTEM3 updated from Brohan et al. (2006). The smooth curves show decadal variations … The black curve from CRUTEM3 is compared with those from NCDC (Smith and

Reynolds, 2005; blue), GISS (Hansen et al., 2001; red) and Lugina et al. (2005; green). See http://

www.ipcc.ch/

(5)

ISI  Congress,  Dublin,  August  26th  2011   5  

Why  the  need  to  revisit?  

Global  surface  temperature  records  are  key  line  of  evidence  

Backed  by  many  other  indicators  (ice  loss,  humidity  increases,  

sea  level  increases  etc.)  

BUT:  

Link  from  sta6on  data  to  global  averages  is  indirect  (sta6s6cal  

problem  –  reconstruc6on  depends  on  methodology  used)  

Data  holdings  are  dispersed  with  poor  provenance    

Few  user  tools:  access  difficult  for  non-­‐specialists  

Concerns  over:  evolving  sta6on  networks;  changes  in  

measurement  technique  /  sta6on  loca6on  /  recording  

prac6ce  (oYen  undocumented);  urban  “heat  island”  effects;  

etc.  etc.    

Increasing  role  of  “blogosphere”  &  cri6ques  outside  peer-­‐

(6)

Inhomogenei$es:  annual  mean  min  temp  

at  Reno,  Nevada,  USA  

(7)

Issues  not  well  understood  …  

 Not  all  commentators    /  policymakers  understand  the  

issues:  

(8)

The  big  ques$on  

Can  we  create  a  process  that  leads  to  a  suite  of  well  

characterized  es6mates  of  land  surface  temperatures  

that  can  be  used  to  answer  scien6fic  and  societal  

demands  of  the  21

st

 Century?  

Open  and  transparent  

Becer  understanding  of  fundamental  instrument  

performance  and  measurement  proper6es  

Consistent  performance  evalua6on  

User  tools  

Not  just  monthly  at  the  largest  scales.  Daily,  sub-­‐daily,  

(9)

How  the  ini$a$ve  was  started  

2010  UK  Met  Office  Submission  to  World  Meteorological  

Organiza6on  Commission  for  Climatology  

–  Call  for  crea6ng  new  suite  of  products  to  meet  21st  Century  

demands  /  expecta6ons  

• 

September  2010  ins6ga6on  workshop,  UK  Met  Office,  

Exeter  UK  

–  80  interna6onal  experts  including  climate  scien6sts,  

metrologists,  sta6s6cians,  soYware  engineers  

–  White  papers  posted  online  and  public  comments  solicited  

–  Agreed  project  outline  and  governance  structure  

–  Agreed  outcomes  published  in  Bull.  Amer.  Met.  Soc.  doi:  

(10)
(11)

Databank  Progress  

Working  Group  ins6gated  

Data  rescue  task  team  

Provenance  and  version  control  task  team  

Development  version  posted  

hcp://www.gosic.org/

GLOBAL_SURFACE_DATABANK/GBD.html  

First  version  release  and  accompanying  

(12)

Stage  0  data  in  hardcopy  /  image  

format  

Millions  of  rescued  images  in  the  NOAA  

Foreign  Data  Library  

2000+  boxes  of  data  in  the  NCDC  library  

Holdings  in  other  libraries  and  repositories,  

par6cularly  former  colonial  powers  

Holdings  literally  ropng  away  or  seen  as  a  

nuisance  in  many  countries  

(13)
(14)
(15)
(16)
(17)

Data  currently  digi$zed  

Numerous  public  holdings  

NCDC  

UCAR  

RIHMI  

Regional  and  na6onal  holdings  

Some  restricted  for  commercial  or  geopoli6cal  

(18)

Stage  2  –  common  format  

(19)

Stage  3  

Same  format  as  stage  2  

One  unique  version  for  each  sta6on  –  

recommended  version  for  most  users  

Forms  basis  for  crea6on  of  analogs  (see  later)  

Protocols  used  in  merging  sources  are  as  yet  to  

be  finalized  

Provenance  tracking  will  ensure  an  unbroken  

chain  to  earlier  stages  

(20)

Partnerships  essen$al  

•  Bring  together  exis6ng  efforts,  augment  and  ensure  pull  through.  e.g.  ACRE     ( Atmospheric  Circula6on  Reconstruc6ons  over  the  Earth )  project  (hcp://

www.met-­‐acre.org/),  Interna6onal  Environmental  Data  Rescue  Organisa6on  

(www.iedro.org)  and  other  na6onal  /  interna6onal  programs.  

•  Pursue  innova6ve  approaches  (crowdsourcing  building  upon  success  of   oldweather.org  etc.)  

•  Build  on  ICOADS  model  for  sea  surface  temperatures  (hcp://

icoads.noaa.gov/)  –  easy  submission  and  access  to  data  to  recognized  global  

repository  

•  Recognize  key  partners  and  contribu6ons  

•  And  you  can  also  help  …  

–  www.surfacetemperatures.org/databank  

–  [email protected]  

–  hcp://www.surfacetemperatures.org/databank/DataSubmission-­‐Stage1-­‐Guidance.pdf?

acredirects=0  

(21)

Mul$plicity  of  data  products  

Structural  uncertainty  is  the  key  

–  Raw  data  is  far  from  traceable  to  interna6onal  measurement  

standards  

–  Data  artefacts  are  numerous  and  have  myriad  causes  

–  Metadata  is  patchy  at  best  

–  Data  is  discrete  in  both  space  and  6me  

–  No  “how  to”  …  rather  very  many  cases  of  “it  may  work  …”  

–  Mul6ple  subjec6ve  decisions  required  even  in  automated  

procedures  (thresholds,  periods,  test  type  etc.)  

–  Different  approaches  may  have  different  strengths  and  

weaknesses  

(22)

•  Need  mul6ple  independent  efforts  with  different  choices  

–  Sta6on  selec6on  

–  Time  and  space  resolu6on  

–  Quality  control  choices  

–  Homogeniza6on  decisions  

–  Averaging  procedures    

•  Should  not  just  be  climate  scien6sts  as  need  broad  range  of  

approaches  

–  Sta6s6cians,  metrologists,  soYware  engineers,  ci6zen  scien6sts  etc.  

•  Dis6nct  approaches  pinpoint  key  uncertain6es  so  redundancy  is  

of  fundamental  scien6fic  value  …  

•  …  at  least  providing  the  underlying  algorithm  is  understood  

(23)

Benchmarking  and  assessment  

With  real  world  data  we  do  not  have  the  luxury  of  

knowing  the  truth  –  we  cannot  measure  

performance  of  a  specific  method  or  closeness  to  

real  world  truth  of  any  one  data-­‐product.    

Instead  focus  on  performance  of  underlying  

algorithms  

Consistent  synthe6c  test  cases,  simula6ng  real  world  

(24)

Benchmarking  progress  

Working  group  and  task  teams  ins6gated  

Create  c.10  analog-­‐error-­‐worlds  

–  Climate  model  basis  (maintains  plausible  far  field  correla6on  

structure)  tweaked  with  real  sta6on  climate  characteris6cs  

–  Add  in  random  and  systema6c  errors  to  approximate  the  real  

world  error  structures  which  may  exist  

–  Algorithm  assessment  based  upon  ability  to  recover  original  

data  

–  Error  structures  should  enable  answering  a  range  of  ques6ons  /  

assump6ons  regarding  the  true  error  to  avoid  over-­‐tuning  

Analogs  to  be  made  available  Nov.  2012  based  upon  version  1  

(25)
(26)

Benchmarking  example  

For  USHCN  (lower  48  states)  

Step  changes  and  white  noise    

–  4  iden6cal  error  structure  on  2  20th  Century  runs,  control  and  

emission  halt  runs  from  different  models  

–  No  errors  (do  no  harm)  

–  Predominantly  frequent  small  breaks  (hard)  

–  Perfect  metadata,  large  breaks  (easy)  

100  member  perturbed  ensemble  of  the  NCDC  pairwise  

algorithm  

Considera6on  solely  of  6meseries  and  trends  

Analyses  that  follow  are  for  predominantly  frequent  small  

(27)
(28)
(29)
(30)

Area  averaged  $meseries  and  trends  

(31)
(32)

Open  issues  on  data  product  

provision  

Data  formats  

Degree  of  user  interac6on  

Ability  to  create  graphical  and  tabular  output  

on  the  fly  

Limited  progress  to  date  

Largely  a  reflec6on  that  this  data  provision  is  

some  way  down  the  road?  

(33)

Progress  to  date  summary  

Steering  commicee  set  up  

Terms  of  reference  

World  Meteorological  Office  and  The  Interna6onal  

Environmetrics  Society  sponsored  (seeking  Interna6onal  

Bureau  of  weights  and  measures)  

Working  groups  on  databank  and  benchmarking  ac6ve  

Databank  prototype  made  public  and  data  sources  coming  

in.  

Implementa6on  Plan  published  

Progress  documented  on  ini6a6ve  website  at  

(34)

You  can  help  make  this  a  success  

Help  to  find  raw  data  sources  

Come  up  with  novel  ways  of  analyzing  the  

data  

Partake  in  the  benchmarking  exercise  

Help  to  make  a  data  portal  a  reality  

(35)

Ques$ons  and  Answers  

www.surfacetemperatures.org    

[email protected]    

(36)

Background  materials  

Links  to  other  relevant  organiza6ons  (finite)  

hcp://www.met-­‐acre.org/

   -­‐  ACRE  –  data  rescue,  

reanalyses,  end-­‐user  engagement  

hcp://iedro.org/

 -­‐  IEDRO  –  data  rescue  

hcp://icoads.noaa.gov/

 -­‐  ICOADS  –  marine  

databank  

hcp://www.data-­‐rescue-­‐at-­‐home.org/

 -­‐  data  

(37)

Ecological Phenological Health & Disease Reinsurance Climate Monitoring Model Validation Environmental

Assessments Extremes, Impacts & Risks Water resources Agricultural Forestry Energy Marine operations Fisheries Cultural landscapes and

built heritage Education Global historical reanalysis

56 realisations every 6 hours, 2⁰ x 2 ⁰ spatial resolution

O p en Pl atfo rm Me t O ffi ce -T SB -I B M PR EC IS D ow nsca lin g H ig he r R eso lu tio n Citizen Science Crowd Sourcing

(eg. Oldweather.org

Data rescue@home)

Global surface weather observations over the last

References

Related documents