16
GPM Applications WS Weather Forecasting PB 11/2013 ECMWF GPM and Weather Forecas.ng Peter Bauer E uropean C entre for M ediumrange W eather F orecasts Reading, UK

GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

GPM    and  

Weather  Forecas.ng    Peter  Bauer    European  Centre  for  Medium-­‐range  Weather  Forecasts    Reading,  UK  

Page 2: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

•  An  independent  intergovernmental  organisa.on;  established  in  1975  •  20  Member  States,  14  Co-­‐opera.ng  States,  45  M£  annual  budget  (staff:HPC)  •  270  staff  (110  RD,  55  CD,  65  FD,  40  AD)  

•  No  regional  systems  •  No  warnings  issued  •  Only  limited  value  adding  applied  to  forecast  model  output  

ECMWF  

L137/0.01  

L137/0.01  

L137/0.01  

L91/0.01  

L137/0.01  

L91/0.01  

L60/0.1  

Page 3: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

Assimilated  satellite  data  

…  and  GPM  

2013:  50+  instruments  assimilated,  ~107  observaDons  per  day  

Page 4: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

Data  assimilaDon  system  

Analysis  =  ini.al  state          Forecast  

Page 5: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

Data  assimilaDon  system  

•  95%  of  all  data  is  assimilated  as  radiances  (i.e.  level-­‐1)  •  100%  of  cloud/precipita.on  satellite  data  is  assimilated  as  radiances  (i.e.  level-­‐1*)  •  Cloud/precipita.on  assimila.on:  

•  Opera.onal  since  2005  with  major  upgrades  in  2009,  2011,  2013  •  8  years  of  development  .me  x  2-­‐3  FTE  

 (*  with  SSM/I,  TMI,  SSMIS,  ASMR-­‐E,  AMSR2)  

Analysis  =  ini.al  state          Forecast  

Page 6: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

VerificaDon  of  model  forecasts  with  TRMM  TRMM            

Old  model          

 New  model  

Page 7: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

VerificaDon  of  model  forecasts  with  TRMM  TRMM            

Old  model          

 New  model  

TRMM      

     Old  model          

 New  model  

Page 8: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

•  AMSU-­‐A  currently  most  important  single  observing  system  (5  satellites)  =  fct.  (data  volume,  single  observa.on  impact,  synergy  with  others)  

•  Microwave  imagers  most  important  instrument  for  lower  tropospheric  moisture!  

Data  impact  on  forecasts:  Average  

Page 9: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

FEC:  Log-­‐scale  

MSLP  

All  observa.ons  from  which  Forecast  Error  Reduc.on  <  -­‐20  kJ/kg  

Data  impact  on  forecasts:  Case  North  AtlanDc  storm  

Page 10: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

MSLP  

FEC-­‐contribu.on  by  type  for  analysis  6  November  00  UTC  in  30ox30o  box  

Data  impact  on  forecasts:  Case  North  AtlanDc  storm  

Page 11: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

Sandy  

hfp://intra.ecmwf.int/publica.ons/cms/get/weekly_news/2012-­‐11-­‐02  

Forecast  24/10/2012  +144h  

Page 12: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

Data  impact  on  forecasts:  Sandy  

Full  EOS  forecast   Verifying  analysis  No  MW  imagers  

No  MW/IR  sounders  

No  LEO  satellites   No  GEO  satellites   No  drop  sondes  

No  GPS-­‐RO            No  AMV            

Page 13: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

Model  impact  on  forecasts:  Sandy  

Radar  20121030  12  UTC  

120  km  

5  km  16  km  

Page 14: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

NWP  systems:  •  Improved  hybrid  data  assimila.on  to  produce  op.mal  analysis  (ini.al  state)  +  uncertain.es  •  Resolu.ons  of  10  km  in  2015,  5  km  in  2020,  2.5  km  in  2025  •  Coupling  with  ocean/sea-­‐ice/atmospheric  composi.on  (model  and  assimila.on)  •  Befer  characterisa.on  of  observa.on  +  model  errors  •  Computer  codes  that  scale  on  massively  parallel  HPC  

Future  

Page 15: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

NWP  systems:  •  Improved  hybrid  data  assimila.on  to  produce  op.mal  analysis  (ini.al  state)  +  uncertain.es  •  Resolu.ons  of  10  km  in  2015,  5  km  in  2020,  2.5  km  in  2025  •  Coupling  with  ocean/sea-­‐ice/atmospheric  composi.on  (model  and  assimila.on)  •  Befer  characterisa.on  of  observa.on  +  model  errors  •  Computer  codes  that  scale  on  massively  parallel  HPC    Satellite  data:  •  Bulk  of  data  as  level-­‐1  product  (radiance,  reflec.vity,  backscafer  x-­‐sec.on)  •  Increased  usage  of:  

•  Cloud/precipita.on/water  vapour  •  Aerosol/composi.on    •  Land  surface          •  Snow/ice  

Future  

Page 16: GPM$ · GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF GPM$ and$ Weather$Forecas.ng$ $ Peter$Bauer$ $ European$Centre$for$Medium8range$

GPM Applications WS Weather Forecasting PB 11/2013 Ⓒ ECMWF

NWP  systems:  •  Improved  hybrid  data  assimila.on  to  produce  op.mal  analysis  (ini.al  state)  +  uncertain.es  •  Resolu.ons  of  10  km  in  2015,  5  km  in  2020,  2.5  km  in  2025  •  Coupling  with  ocean/sea-­‐ice/atmospheric  composi.on  (model  and  assimila.on)  •  Befer  characterisa.on  of  observa.on  +  model  errors  •  Computer  codes  that  scale  on  massively  parallel  HPC    Satellite  data:  •  Bulk  of  data  as  level-­‐1  product  (radiance,  reflec.vity,  backscafer  x-­‐sec.on)  •  Increased  usage  of:  

•  Cloud/precipita.on/water  vapour  •  Aerosol/composi.on    •  Land  surface          •  Snow/ice  

Requirements:  •  Con.nuity  of  core  observa.on  types  (sounders,  imagers)  with  sufficient  global  coverage  •  Enhanced  modelling  capabili.es  for  process  representa.on  •  Enhanced  observaDonal  capabili.es  for  process  studies  and  verifica.on  •  Enhanced  data  assimilaDon  capabili.es  for  op.mal  data  exploita.on  

→  Only  combinaDon  of  the  above  will  produce  return  on  investment!  

Future