We think you have liked this presentation. If you wish to download it, please recommend it to your friends in any social system. Share buttons are a little bit lower. Thank you!
Presentation is loading. Please wait.
supports HTML5 video
Published byChasity Fickling
Modified over 3 years ago
NWP in the Met Office © Crown copyright 2006
Topics to be covered... 1. Describing the atmosphere2. Using observations 3. Model mathematics 4. Operational models purposes 5. Model outputs © Crown copyright 2006
The Weather Prediction ProcessOBSERVATIONS NUMERICAL FORECASTS R&D VERIFICATION CUSTOMERS ARCHIVES Observations are made at set times of the day and increasingly in a continuous manner. Collection of observations from around the world + 5 day fc takes < 4hrs Use the CRAYT3E to produce the forecasts Many customers take output without human intervention eg aviation HUMAN FORECASTER © Crown copyright 2006
Unified Model Met. Office has several requirements: local forecastingglobal forecasting climate modelling ocean and wave modelling a common model: shares code and operating structure is modular where differences are necessary gives considerable savings in maintenance cost © Crown copyright 2006
Unified Model configurations© Crown copyright 2006
Met Office models Based on Unified Model GlobalNorth Atlantic and European model 4km and 12km Mesoscales Crisis Area Mesoscale Models Stratospheric FOAM ocean forecasting models © Crown copyright 2006
Met Office models Other models includingWave (Global, European, UK Waters) Surge NAME SSFM © Crown copyright 2006
Fundamentals of NWP 1. Specify atmospheric initial conditions in a numerical form 2. Use equations describing atmospheric physical processes to predict how the initial state will evolve 3. Output the forecast in a useful form for the user © Crown copyright 2006
1. Describing the atmosphere
Unified Model informationUnified Model is a grid point model Non-hydrostatic Semi-Lagrangian advection Semi-implicit time integration 20 min timestep: Global 5 min timestep: Mes and NAE Horizontal staggering - Arakawa C grid Vertical staggering - Charney – Philips Uses 4-D Var data assimilation © Crown copyright 2006
Specify the properties in the grid box Unified Model is a gridpoint model Grid length Grid point Specify the properties in the grid box from observational data (temp, pressure humidity, wind etc.) © Crown copyright 2006
Vertical co-ordinates in the UMHybrid height (z) co-ordinates At the model top =1 so z=H Above the first flat eta level =z/H z= H I H z H Below the first flat eta level (I) z= h(1- /I)2 where h=orography height h At the surface = 0 so z=0 © Crown copyright 2006
GM Vertical resolution – 50 levels65 km In free atmosphere levels are height coordinates In between levels are a combination of the 2 17.5 km Lowest model levels present/new 70L at 10m/2.5m for wind at 20m/5m for temp The 50 level model has extra levels in stratosphere (The same resolution below). The business case for this is to: improve the analysis of satellite radiance channels, which peak in the stratosphere but have a broad weighting function covering including the significant tropopause levels. move the model upper lid further away from the areas of significant meteorological interest. The upper lid is likely to be at 65km rather than 40km. have a single global model covering all customer interests and are able to pension off the separate stratospheric model and reduce maintenance costs. In a 70 level model, lowest model levels are at 5m for temperature and 2.5 metres for wind, : In boundary layer levels are terrain-following © Crown copyright 2006
Global, North Atlantic & European, mesoscale modelsGlobal Model (GM) Horizontal Resolution: Mid-latitude 40km Timestep: 20mins Vertical levels: 50, then 70 Grid: Standard lat/long type, with filtering near the poles North Atlantic & European Model (NAE) Horizontal Resolution: 12km Timestep: ~5 mins Vertical levels: 38, eventually 70 Grid: Rotated lat/long (‘Equatorial Lat-long Fine-mesh’ - ELF) Mesoscale Model (MES) Horizontal Resolution: 12km/4km Timestep: 5/ 1.7 mins Vertical levels: 38, eventually 70 Grid: Rotated lat/long (‘Equatorial Lat-long Fine-mesh’ - ELF) Extra resolution in the boundary layer (when go to 70 levels) The new model will have twice as many levels in the lowest kilometre. (14 compared with 7) The lowest model levels will be 5m for temperature (currently 20m) and 2.5m for wind (currently 10m). This option provides much better chance of resolving thin fog and stratus layers. It is comparable to the vertical resolutions used by the site specific forecast model so should provide the same benefits. Extra resolution in mid tropospheric levels Between 800hPa and 400hPa, the new model will have twice as many levels (24 compared with 12) The coarsest resolution in terms of pressure is 22hPa between the levels at 500hPa compared with 45hPa at present The resolution increase at the mid tropospheric levels provides should benefit cloud and precipitation. Extra resolution at jet levels At present there are 3 levels between 300hPa and 200hPa. In the new model we will 5 levels. The spacing between levels reduces from 1km to 500m (or from about 30hPa to 20hPa). This should reduce model errors at this meteorologically significant level and should improve the model accuracy generally. It also of course has a direct benefit to aviation. © Crown copyright 2006
2. Using observations
Data assimilation GM uses 4-D VAR; 12km MES and NAE 3-D VAR4km MES has no data assimilation yet Model is run for an assimilation period prior to the forecast 6 hrs for GM model 3 hrs for the MES and NAE © Crown copyright 2006
Data assimilation Observations firstly quality controlled againstclimate data model background field nearby obs. Then inserted into the run at or near their validity time to nudge the model towards reality © Crown copyright 2006
Using observations Models try to make the best possible use of observations Observations are checked for quality and interpolated onto the model grid points Different types of data have different areas of influence airep GP GP GP sonde sonde ship synop synop synop LAND synop GP GP GP ship synop SEA GP GP GP © Crown copyright 2006
3- and 4-Dimensional VAR 3DVAR 4DVARminimises the equation for a given time (e.g. 12Z) forms an analysis at a point in time forms an analysis which is consistent with a static state of the atmosphere MES T+12 and T+24 forecasts differences verifying at same time give error characteristics of model (8 months of cases used) 4DVAR minimises the equation by running the model backwards and forwards over an analysis period (e.g. 6 hours) forms an analysis for a period forms an analysis which is consistent with dynamical evolution of the atmosphere © Crown copyright 2006
Moisture Observation Pre-processing System (MOPS)Used only in 12km MES/NAE Latent heating and cooling important in driving mesoscale systems MOPS is an analysis of humidity, cloud and precipitation for 12km MES and NAE © Crown copyright 2006
Soil moisture in the GM No longer reset weekly to climatologyNew soil moisture nudging scheme Not as complex as MOPS Produced verifiable improvement, especially surface temperatures © Crown copyright 2006
3. Model Mathematics
Model variables PRIMARY PROGNOSTIC variables are explicitly calculated using the primitive equations ANCILLARY FIELDS are fixed lower boundary conditions SECONDARY PROGNOSTIC variables are calculated at each timestep from the prognostic variables. © Crown copyright 2006
Primary prognostic variablesHorizontal and vertical wind components potential temperature specific humidity cloud water and ice surface pressure surface temperature soil temperature canopy water content snow depth © Crown copyright 2006
Ancillary fields land/sea mask soil type vegetation type grid-box mean and variance of orography sea surface temperature proportion of sea-ice cover sea-ice thickness sea surface currents Prognostic variables in coupled atmosphere/ocean models © Crown copyright 2006
Global model orography© Crown copyright 2006
NAE Model orography © Crown copyright 2006
12km / 4km MES Model orography© Crown copyright 2006
New UK 4 km Model Broad Leaf Trees Needle Leaf Trees C3 Grass C4 GrassShrubs Urban Lakes Bare Soil Land Ice © Crown copyright 2006
Model variables PRIMARY PROGNOSTIC variables are explicitly calculated using the primitive equations SECONDARY PROGNOSTIC variables are calculated by the parameterisation schemes © Crown copyright 2006
Model variables primary prognostic variableshorizontal and vertical wind components potential temperature specific humidity cloud water and ice surface pressure surface temperature soil temperature canopy water content snow depth secondary prognostic variables boundary layer depth sea surface roughness convective cloud amount convective cloud base convective cloud top layer cloud amount ozone mixing ratio © Crown copyright 2006
Parametrised processes1. Layer cloud and precipitation 2. Convective cloud and precipitation 3. Radiative processes 4. Surface and sub-surface processes 5. Gravity wave drag © Crown copyright 2006
1. Layer cloud and precipitation* * * * * * * * * * * * * © Crown copyright 2006
Convective cloud model2. Convective cloud and precipitation Convective cloud model © Crown copyright 2006
3. Radiative processes © Crown copyright 2006
4. Surface and sub-surface processes© Crown copyright 2006
5. Gravity wave drag © Crown copyright 2006
Boundary conditions Lower and upper boundaries Lateral boundariesLand & sea: ancillary fields Stratosphere: ‘lid’ to model required in MES and NAE models primary prognostic variables required at each grid point NAE and 12km MES supplied from global model 4km MES supplied from NAE possible source of error © Crown copyright 2006
4. Purposes of Operational Models
Global model 4 times daily Run times … 00Z, 06Z,12Z, 18ZData accepted up to T+1 hour 45 min Out to T+144 (6 days) at 00Z and 12Z, T+48 at 06Z and 18Z Takes 2hr 15 mins to run out to T+144, 1hr 15min for T+48 © Crown copyright 2006
Global model Used for:- regional synoptic guidancemedium range guidance civil aviation products mesoscale model boundary conditions © Crown copyright 2006
North Atlantic & European modelRun times … 00, 06, 12 and 18Z Takes boundary conditions from Global Model (previous GM run) Run partly overlaps with the GM Out to T+48 © Crown copyright 2006
North Atlantic & European modelUsed for:- wider range of products to international customers Improved Synoptic development guidance Better for rapid developments and extremes Boundary conditions for 4km MES © Crown copyright 2006
Advantages of NAE ModelLarge domain Captures developing systems over North Atlantic Covers all of Europe and European Nimrod area includes some other model areas Higher resolution than GM (12km-v-40km) Better for rapid developments and extremes © Crown copyright 2006
12km Mesoscale model Run times … 00Z, 06Z, 12Z and 18ZTakes boundary conditions from Global Model Runs in parallel with the GM (starts 10 mins later) Out to T+48 (2 days) © Crown copyright 2006
12km Mesoscale model Used for:- UK local detail (ppn, cloud,temp,wind)Input to other systems (SSFM, and Nowcasting systems etc.) © Crown copyright 2006
4km MES model Run times … 03Z, 09Z, 15Z, 21ZTakes boundary conditions from NAE Model Out to T+36 No data assimilation © Crown copyright 2006
The Site Specific Forecast Model Philosophyis a fast but comprehensive 1D physical model based on the UM is coupled to NWP output makes timely use of local observations more rapidly than is feasible with NWP adds “local” detail of surface and (limited) orography provides (semi-) automated forecasts of near- surface temperatures, radiation fog, very low cloud (etc.) © Crown copyright 2006
Description & PerformanceThe SSFM 1D model based on UM “physics” - full column! Greatly increased resolution in BL & soil “Dynamics”=“Forcing data”: grad p, advection, etc. Simple forcing correction for orography MOSES with tile surface exchange for separate treatment of land use types Radiative canopy coupled to surface exchange Upwind satellite derived land-use determines drag & surface fluxes of heat, moisture Surface landuse weighting via a stability dependent Source Area Model Description & Performance © Crown copyright 2006
Source Areas Site Wind Source Area Weighting © Crown copyright 2006
SSFM land use and orography100 m resolution orography © Crown copyright 2006
SSFM Applications First guess in Open Road Semi-automatic TAF systemFirst guess for utilities forecasts (METGAS, NGC) Input for air quality forecasts Products for media database © Crown copyright 2006
Model Dependencies (simplified!)NAME GLOBAL GLOBAL WAVE FOAM NAE MODEL EUROPEAN WAVE SHELF SEAS 4KM MES 12KM MES SSFM UK WATERS WAVE SURGE Scheduling must account for all dependencies and timeliness requirements of each model run © Crown copyright 2006
5. Model output
Criteria for allocating symbols on 6-up chartsmm/hr If convective rate reaches 0.5 mm/hr heavy shower symbol is plotted (same for snow) For dynamic rainfall rate of 4 mm/hr heavy rain symbol is plotted If dynamic rainfall rate <0.1 mm/hr type with highest rate is plotted If dynamic rate is >0.1 mm/hr dynamic symbols are plotted © Crown copyright 2006
6-up Frames-problems Grid point averaging Thinned grid Snow prob linesSymbolic representation © Crown copyright 2006
Global model precipitation forecast T+36 © Crown copyright 2006
T+36 forecast rainfall/MSLP Valid 6Z, 1st Dec 2003.Global Model NAE Model © Crown copyright 2006
T+12 forecast rainfall/MSLP Valid 12Z, 1st Jan 2005.NAE MES GM © Crown copyright 2006
T+18 forecast rainfall/MSLP Valid 18Z, 4th September 2005.GM NAE 12km MES 4km MES © Crown copyright 2006
4km Mesoscale model visibility forecast. 9Z 21 Nov© Crown copyright 2006
Site Specific Forecast Model meteogramCloud cross-section Temperature, precipitation, road condition plot Wind plot © Crown copyright 2006
Boscastle, Cornwall. Flooding 16th August 2004RADAR © Crown copyright 2006
Nowcasting Systems in the Met OfficeNimrod- dynamic rain, and other weather elements 0-6 hrs, 5km res Gandolf- convective rain, 0-6 hrs, 2km res Convection Diagnosis Project – Probabilistic convective product hrs, 5km res © Crown copyright 2006
Gandolf Nowcasts from 13UTC 16/8/04T+1: 14UTC T+3: 16 UTC © Crown copyright 2006
6. NWP Verification
Global NWP Index Weightings applied Weightings add to 100Weighted by relative importance to customers 36-month running means © Crown copyright 2006
Global NWP Index © Crown copyright 2006
© Crown copyright 2006
7. Future developments
Future developments 12km N. Atlantic European Model (Mar 2005) Regional Ensemble capability (Aug 2005) Prognostic cloud/condensate scheme (Nov 2005) 45km Global Model (Dec 2005) Regional 4D-Var (Dec 2005) 4km UK Model (Apr 2006) Multi-model global ensembles out to 14-days (Apr 2006) Use of new data from METOP instruments (Sep 2006) Air quality predictions from NWP (Dec 2006) © Crown copyright 2006
Trial hi-res meso model© Crown copyright 2006
Comparison of 00 UTC: 12, 4 and 1 km forecasts. 16 Aug 200412-18 from 00 UTC 1km resolution 12-18 from 00 UTC 12km 12-18 from 00 UTC 4km © Crown copyright 2006
Mechanism for 16th August 2004 Storm Triggering4 km 12 km Persistent Convergence Due to coast and orography 10 m wind convergence at 11 UTC (as convection triggered) © Crown copyright 2006
19Z, 31 Jan 2003- Hi res MES models. T+7 precip & radar12km 4km 1km © Crown copyright 2006
Any questions? GM, NAE and MES output. NWP Gazette NWP technical reports 4km mesoscale runs:
Basics of numerical oceanic and coupled modelling Antonio Navarra Istituto Nazionale di Geofisica e Vulcanologia Italy Simon Mason Scripps Institution.
What’s quasi-equilibrium all about?
Urban Modelling 1 03/2003 © Crown copyright Urban Scale NWP with the Met Office's Unified Model Peter Clark Mesoscale Modelling Group Met Office Joint.
Chapter 13 – Weather Analysis and Forecasting
Weather Research & Forecasting: A General Overview
Page 1 NAE 4DVAR Oct 2006 © Crown copyright 2006 Mark Naylor Data Assimilation, NWP NAE 4D-Var – Testing and Issues EWGLAM/SRNWP meeting Zurich 9 th -12.
Regional Modelling Prepared by C. Tubbs, P. Davies, Met Office UK Revised, delivered by P. Chen, WMO Secretariat SWFDP-Eastern Africa Training Workshop.
Report of the Q2 Short Range QPF Discussion Group Jon Ahlquist Curtis Marshall John McGinley - lead Dan Petersen D. J. Seo Jean Vieux.
PRESENTS: FORECASTING FOR OPERATIONS AND DESIGN February 16 th 2011 – Aberdeen.
Page 1 NAE 4DVAR Mar 2006 © Crown copyright 2006 Bruce Macpherson, Marek Wlasak, Mark Naylor, Richard Renshaw Data Assimilation, NWP Assimilation developments.
© University of Reading 2006www.reading.ac.uk Quasi-stationary Convective Storms in the UK: A Case Study Robert Warren Supervised by Bob Plant, Humphrey.
The Problem of Parameterization in Numerical Models METEO 6030 Xuanli Li University of Utah Department of Meteorology Spring 2005.
1 00/XXXX © Crown copyright Use of radar data in modelling at the Met Office (UK) Bruce Macpherson Mesoscale Assimilation, NWP Met Office EWGLAM / COST-717.
The Forecast Process. Objectives Develop a best practice scheme for the use of data in producing a forecast Produce a ‘flow chart’ illustrating this for.
NOAA/NWS Change to WRF 13 June What’s Happening? WRF replaces the eta as the NAM –NAM is the North American Mesoscale “timeslot” or “Model Run”
For the Lesson: Eta Characteristics, Biases, and Usage December 1998 ETA-32 MODEL CHARACTERISTICS.
GRAPES-Based Nowcasting: System design and Progress Jishan Xue, Hongya Liu and Hu Zhijing Chinese Academy of Meteorological Sciences Toulouse Sept 2005.
03/06/2015 Modelling of regional CO2 balance Tiina Markkanen with Tuula Aalto, Tea Thum, Jouni Susiluoto and Niina Puttonen.
The NCEP operational Climate Forecast System : configuration, products, and plan for the future Hua-Lu Pan Environmental Modeling Center NCEP.
Rapid Update Cycle Model William Sachman and Steven Earle ESC452 - Spring 2006.
© 2018 SlidePlayer.com Inc. All rights reserved.