gnss and weather forecasting - gfg2 · pdf fileprediction model observations ... gps...

© Crown copyright Met Office

GNSS and Weather Forecasting

Dr Jonathan Jones, UK Met Office GfG2 Summer School 2013, GFZ, Potsdam


Contents

•  Meteorological observing systems

•  GNSS-meteorology in Europe

•  Meteorological applications:

•  Numerical Weather Prediction models

•  Forecasters

•  Climate

•  Future developments

Met. Observing Systems

Some of the obs. used in making the analysis: Surface pressure, radiosondes, aircraft data, satellite radiances, satellite clouds and winds, wind profiler radars, buoys, ships, radar data, etc.

Surface pressure observations (land and ships)

Buoy data

Radiosonde launches

Aircraft data - Ascending, descending, and cruising obs. of wind speed, wind direction and temperature

Data assimilation system provide ”Analysis”

(=initial conditions) Numerical weather

prediction model

Observations Boundary values from external

model

Old model state

Computer generated forecast products

Forecasts by forecasters

Non-global NWP System


Global Model:

25km

4D-VAR

NAE Model:

12km

4D-VAR

UK Models:

1.5km

3D-VAR

NWP Models (UK Met Office)

Number of obs. ~10,000 –> 1M Number of model variables ~100M –> 1B We need more observations !!


GNSS-met in Europe

Timeline of European GNSS-Met Projects

1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016

Today

COST Action GNSS4SWEC

E-GVAP

E.C. supported TOUGH Project

COST Action 716

E.C. supported MAGIC Project

E.C. supported WAVEFRONT Project


Current Status in Europe (E-GVAP)

•  Project focusing on GPS-only hourly processing, delivering only Zenith Total Delay (ZTD) in 90mins

•  Operational assimilation at a few Euro National Met Services, many others under testing. Use of ZTD has a positive impact on NWP forecast skill

•  ~1800 European sites’ delivering >12M ZTDs per month

•  Surface T and P used for conversion to Integrated Water Vapour (IWV)

•  GPS IWV has been used in research experiments for more than 10 years

•  Data monitoring and Quality Checking in place (+improving)

Data Quantity

~2350 stations contributing data in NRT



Data Quantity

15M+ ‘humidity’ observations per month ‘free of charge’


Global GNSS Data Coverage


Data Outputs

•  ASCII COST716 format for scientific use

•  Binary ‘BUFR’ format for NWP assimilation

•  Water vapour maps and animations made available to the forecasters and for case studies

•  High quality reprocessed data available for climate applications

Global IWV Maps


Regional IWV Maps


IWV Flux maps



Meteorological Applications


Customers of the Data

•  NWP

Assimilation impact trials show positive impact (in precipitation forecasts and also cloud cover and surface temperature)

•  Forecasters

GNSS WV images give more detail than satellite WV. Useful for forecasters especially in unstable, high humidity convective conditions.

•  Climate

Data potentially of great use to climate community but still in its infancy (+time consuming reprocessing is necessary)


ZTD to ZWD

•  Dry component of delay, ZHD is modelled a-priori in GPS processing using a model such as Saastamoinen:

•  The Zenith Wet Delay (ZWD), is simply the difference between ZTD and ZHD, i.e.

( )rsrs hhPP −−= 119.0

( ) ( )( )[ ]100028.02cos00266.010022768.0 −−−= ss hPZHD φ

ZWDZHDZTD +=


ZWD to IWV

•  If surface temperature is known ZWD can be converted to Integrated Water Vapour (IWV)

•  Which reduces to:

dzTRk

dzRkRkZWDrr z

wwz wdw ∫∫

∞∞ −− +−=ρ

ρ 3612

6 10)(10

IWVTR

kRkRkZWDm

wdw )(10 312

6 +−= −


Data Assimilation into NWP


NWP Observation Operator

∫∞=

=

−=z

z

NdzZTD0

610

2321

Tpk

Tpk

TpkN vdd ++=

2Tbp

TapN v+=

Bevis et al. 1994


Observation Operator

Refractivity

Height


Observation Operator

))(exp( 11 levellevellevellevel zzcNN −−= ++

1

1ln

+

+

−

⎟⎟⎠

⎞⎜⎜⎝

⎛

=levellevel

level

level

zzNN

c

))exp())(exp(exp(10 161

levellevellevellevellevel

level czczczcNZTD −−= +

−+


•  NWP is better at estimating ZHD than ZTD.

•  Assimilation of GNSS ZTD provides information on atmospheric water vapour.

Information Content

NWP ZTD Bias

NWP ZHD Bias


Monitoring


Factors affecting GNSS ZTD Data Assimilation

•  Station lists (whitelist vs. blacklist)

•  Bias Correction (site and AC specific)

•  Height difference correction

•  Spatial and temporal thinning (useful in coarse models and regions where too many obs.)

•  Observation errors


Assimilation Impact Studies


NWP Impact Study, KNMI, 2010

•  Precipitation in KNMI HIRLAM model with no GNSS assimilation



•  Precipitation from weather radar observations



•  Precipitation in KNMI HIRLAM model with GNSS assimilation

•  Note - much better precipitation forecast over Dutch region

37

Impact on Meteo France AROME forecasts

Accumulated rainfall between 03UTC and 15UTC, 19 July 2008

WITHOUT ZTD data assimilation

AROME, 15h forecast starting from the 00UTC analysis, 19 july 2008

WITH ZTD data assimilation

Observations

15/06/2010 – 06UTC

AROME_WMED (D031) OBS (24h accumul.) AROME_WMED (D03Q)

Old white list New white list Rain gauge obs.


GNSS improves location of rainfall

15/06/2010 – 09UTC

OBS (6h Accumul.)

195mm/6hr

AROME_WMED (D03Q)

130mm/6hr New white list

AROME_WMED (D031)

Old white list Rain gauge obs.

GNSS improves severity of rainfall



Latest UK Impact Trials

•  41 day trial – July 2011

•  4km 3D-VAR

Control Trial 1 Trial 2 Trial 3 Assimilation as normal for operational model (observation error = 6mm)

ZTDs not assimilated

ZTD observation error = 9mm

ZTD observation error = 12mm

3.2% increase in Tsurface RMS error




Latest UK Impact Trials

•  6 hour precipitation accumulation

•  Yellow = Trial 1

•  Green = Trial 3

•  Blue = Trial 2

•  Red = Control


Latest Global impact trials

•  41 day Summer trial, 41 day Winter trial

•  40km horizontal resolution

Control Trial 1: Winter 2012 Trial 2: Summer 2012

Assimilation as normal for operational model (observation error = 9mm)

Observation error = 15mm

Observations from GOPG and METG added

Observation error = 15mm

Observations from GOPG and METG added

Zero impact on average 2% decrease in RMS errors across various parameters.

Improvements seen almost entirely in southern hemisphere and tropics


Latest Global Impact Trials

Standard deviation of the difference in ZTD over ten months

σGOPG ZTD - NWP ZTD and σGOPG ZTD - IGS REPRO ZTD


GNSS and Forecaster Trials


24th June 2005 Case Study

•  Synoptically forced event with wide spread thunderstorms throughout Southern UK and the Midlands, flash flooding etc

•  Trough progressing over southern UK from westerly direction with associated high IWV and convective thunderstorm cells


Weather radar

•  Traditional image to forecaster showing line of storms progressing over Southern UK


GPS Water Vapour Animation

•  GPS IWV identified ‘cold pool’ behind front which perpetuated instability and aided further convection


28th July 2005 Case Study

•  Low pressure system centred over S. Irish Sea

•  Winds generally southerly

•  High IWV and tornado in Birmingham at ~13:15

•  High areas of IWV tracks East (not with prevailing wind direction) and second tornado in Peterborough at ~17:00 under sharp IWV gradient



6.2micron Meteosat 8 water vapour image, 18:00UTC 28th July 2005

GPS Water Vapour plot, 17:30, 28th July 2005. White line on plot is manually drawn extent of dry intrusion

© Crown copyright Met Office Drier Upper Air


Dry tongue over South East not corresponding with satellite WV By combining satellite and ground based GPS water vapour you can infer dry intrusion is slanting giving added value to forecasters


ZTD O-B Case Study (16/02/10)


ZTD O-B Case Study 13/10/10


Future Developments


New Meteo Requirements

•  New hi-res NWP models require ZTD with improved timeliness and greater spatial and temporal resolution (e.g. Met Office UKV 1.5km)

•  Advanced GNSS products desired for obtaining more information about troposphere (vertical resolution of water vapour, azimuthal anisotropy etc.)

•  Real-time processing would greatly increase the usefulness of GNSS products for nowcasting.

•  Climate community only now starting to use GNSS tropospheric products (e.g. Hadley Centre)


Recent GNSS Developments

•  More GNSS constellations (GPS + GLONASS, Galileo etc…) = new geometries, increased number of observations

•  Real-time raw data streaming

•  R&D advanced tropospheric products (slants, gradients, tomography etc.)

•  Single frequency processing

•  Long-term, homogenised GPS products available (IGS/EPN/CODE/others), valuable for climate analysis ?

Slants, tomography, RT processing


ES1206: Advanced GNSS Tropospheric Products for monitoring Severe Weather Events and Climate


•  4 year COST Action May 2013 – May 2017

•  25 COST countries participating (+5 non-COST)

•  80+ participants from 50+ institutes, 3 WGs

•  COST supports:

•  Management Committee and Working Group meetings

•  Scientific workshops

•  Short Term Scientific Missions

•  Training Schools

•  Publications, website, public outreach


Main Objectives of GNSS4SWEC

•  Coordinate the development of new multi-GNSS solutions and assess the benefit to meteorology and climate analysis

•  Assess the potential of new GNSS products for use in nowcasting and rapid cycle NWP

•  Determine the added value of the re-processed GNSS tropospheric data to the current state-of-the-art climate research

•  Establish a database of reference tropospheric solutions to validate reprocessed GNSS ZTD/IWV against climate quality data from a range of other instrumentation

•  Stimulate the exploitation of atmospheric data as an input to improve Real-Time GNSS positioning and navigation

•  Standardize the conversion of ZTD to IWV

•  Stimulate exchange of data and expertise in the field of GNSS Meteorology


Questions and answers [email protected] http://egvap.dmi.dk/

http://www.cost.eu/domains_actions/essem/Actions/ES1206

gnss and weather forecasting - gfg2 · pdf fileprediction model observations ... gps...

Documents