Progress Report on the Global Data Processing System, 2000
United Kingdom
Met Office (Bracknell)
1. Summary of highlights
1.1 Forecast models
The main change to the global versions of the Unified Model in the numerical weather prediction suite was the following:
27th June (from 12Z) Introduction of a new land surface exchange scheme, MOSES, including a thermal vegetative canopy, as already used in the UK mesoscale model.
The main change to the mesoscale version of the Unified Model in the numerical weather prediction suite was the following:
6th June 2000 (from 12Z) The parameterisation of the radiative effects of horizontal inhomogeneities in the cloud water path was changed to give better simulations of stratus/stratocumulus clouds.
The main changes to the ocean and waves models during 2000 were:
28th March Introduction of UK waters wave model forecast to T+48
27 June Introduction of UK waters wave model forecast to T+120 using global NWP winds
13 June Introduction of baroclinic shelf seas model of NW European shelf
19th September Upgrade to FOAM
24th October Wave models move to use 10m winds.
The main changes to the Nimrod nowcasting system were as follows:
14th September A wind and pressure forecast were added to Nimrod.
1.2 Observations, quality control and assimilation
Routine monitoring provides regular revisions to acceptance lists for Synops, Aircraft and Sondes. In addition, the following major changes to the system were made:
December 20 (1999) ERS2 scatterometer winds disabled
March 7 Separate mesoscale sea surface temperature analysis introduced, using high resolution satellite data.
March 24 We revised the MOPS cloud analysis to use less satellite data at high levels in the mesoscale assimilation system.
May 17 Global data assimilation spring upgrade package :
June 6 Mesoscale data assimilation spring upgrade package:
September 26 Started to use radar weight data in MOPS (Moisture Observation Pre-processing System) rain-rate assimilation
November 14 Converted stratospheric data assimilation system to 3DVAR
2. Equipment in use at the centre
A) Front end mainframe computers B) Supercomputers
2.1.1 Make and model of computer
A) IBM 9672 – R45 B) Cray T3Ea (880 PEs)
IBM 9672 – R25 Cray T3Eb (640 PEs)
(PE – Processor Element)
2.1.2 Main Storage
A) 2 Gbytes (R45) B) 128Mb per PE (T3Ea)
1 Gbyte (R25) 256Mb per PE (T3Eb)
(16 PEs on each system have 512Mb)
2.1.3 Operating system
A) OS/390 Version 2 Release 9 B) UNICOS/mk 2.0.5
2.1.4 External input/output devices
A) 1 Terabyte of online disk storage B) 1440 Gbytes (T3Ea)
LAN attached Desktop PCs, 1920 Gbytes (T3Eb)
Workstations and printers
2 line printers
1 microfiche processor
32 magnetic cartridge drives connected
to a GRAU automated tape library system
with a capacity of 28,800 cartridges.
18 StorageTek 9840 cartridge drives
connected to a StorageTek Powderhorn tape
library with a capacity of 5700 cartridge slots.
FileTek StoreHouse data server (MASS)
Sun E6500 UltraSPARC processor and 16
StorageTek 9840 cartridge drives connected
to a StorageTek Powderhorn tape library with
a capacity of 5700 cartridge slots giving a total
volume storage of 78 TB.
2.2 Desktop systems for forecasters
The workstation-based ‘Horace’ system is used for visualisation and production and is operational in the National Meteorological Centre (NMC), Bracknell, and at other major operational locations in the UK.
Each user site comprises at least one Hewlett Packard UNIX data server plus as many multi-screen workstations, printers and plotters as are necessary to meet the local requirements. Communications services via a message switch provide every type of observational data from the GTS, while an ftp server provides the imagery, rainfall and numerical weather prediction (NWP) files.
Horace can display a wide variety of information in many forms, typically a combination of observations, NWP data or imagery (radar and satellite). In addition to its powerful visualisation capability Horace provides many semi-automated facilities and production tools:
3. Data and products from GTS in use
3.1 Observations
The global data assimilation system makes use of the following observation types. The counts are typical of late October, excluding data received but not yet processed for assimilation.
Observation Group |
Observation Sub-group |
Items used |
Daily Extracted |
% used in assimilation |
Ground-based vertical profiles |
TEMP |
T, V, RH processed to model layer average |
1200 |
97 |
PILOT |
As TEMP but V only |
900 |
99 |
|
PROFILER |
As TEMP but V only |
300 |
0 |
|
Satellite-based vertical profiles |
TOVS |
Radiances directly assimilated with channel selection dependent on surface, instrument and cloudiness |
54000 |
11 |
ATOVS |
700000 |
4 |
||
Aircraft |
Manual AIREPS |
T, V as reported with duplicate checking and blacklist |
14000 |
21 |
|
Automated ACARS/ AMDAR/ ASDAR |
|
67000 |
60 |
Satellite atmospheric motion vectors |
GOES 8,10 |
High res. 'BUFR' IR winds |
55000 |
24 |
Meteosat 5,7 |
IR, VIS and WV winds |
9200 |
98 |
|
GMS 5 |
IR, VIS and WV winds |
5200 |
93 |
|
Satellite-based surface |
ERS-2 |
In-house wind vector retrieval |
170000 |
0 |
SSMI-13 |
In-house 1DVAR wind speed retrieval (no moisture yet) |
1450000 |
1 |
|
Ground-based surface |
Land Synop |
Pressure only (processed to model surface) |
27000 |
80 |
Ship Synop |
Pressure and Wind |
6000 |
90,95 |
|
Buoy |
Pressure |
9000 |
75 |
3.2 Gridded Products
Products from WMC Washington are used as backup in the event of a systems failure (see section 7.2.3). The WAFS Thinned GRIB products at an effective resolution of 140 km (1.25º x 1.25º degrees at the equator) are received over cable in 6 hour intervals out to T+72. Since October 1996 we have also been receiving these products over the ISCS satellite link. Fields in this format include geopotential height, temperature, relative humidity, horizontal and vertical components of wind on most of the standard pressure levels, rainfall, PMSL and absolute vorticity.
Products received from Météo France, DWD and ECMWF (including Ensemble Prediction System forecasts) are used internally for national forecasting.
4. Data input system
Fully automated.
5. Quality control system
5.1 Quality control of data prior to transmission on the GTS
Both manual and automatic checks are performed in real-time for surface and upper-air data from the UK, Ireland, Netherlands, Greenland and Iceland. Checks are made for missing or late bulletins or observations, and incorrect telecommunications format. Obvious errors in an Abbreviated Heading Line are corrected before transmission onto the GTS.
5.2 Quality control of data prior to use in numerical weather prediction
All conventional observations (aircraft, surface, radiosonde and also atmospheric motion winds) used in NWP pass through the following quality control steps:
Failure at step 1 is fatal, and the report will not be used. The results of all the remaining checks are combined using Bayesian probability methods (Lorenc and Hammon, 1988). Observations are assumed to have either normal (Gaussian) errors, or gross errors. The probability of gross error is updated at each step of the quality control, and where the final probability exceeds 50 per cent the observation is flagged and excluded from use in the data assimilation.
Special quality control measures are used for satellite data according to the known characteristics of the instruments. For instance, ATOVS radiance quality control includes a cloud and rain check using information from some channels to assess the validity of other channels (English et al, 2000)
6. Monitoring of the observing system
Non-real-time monitoring of the global observing system includes:
Within the NWP system, monitoring of the global observing system includes:
7. Forecasting system
The forecasting system consists of:
1. Global atmospheric data assimilation system (3DVAR)
2. Global atmospheric forecast model
3. Mesoscale atmospheric data assimilation system (3DVAR)
4. Mesoscale atmospheric forecast model
5. Stratospheric global atmospheric data assimilation system (3DVAR)
6. Stratospheric global atmospheric forecast model
7. Transport and dispersion model
8. Nowcasting model
9. Global wave hindcast and assimilation/forecast system
10. Regional wave hindcast and forecast system
11. Mesoscale wave hindcast and forecast system
12. Mesoscale model for sea-surge.
13. Global ocean model.
The global atmospheric model runs with 3 different data cut-off times:
The latest update run provides initial starting conditions for both the early preliminary and main runs of the global atmospheric model. The global atmospheric model provides surface boundary conditions for the global wave and ocean models. The preliminary global provides lateral boundary conditions for the mesoscale model, and surface boundary conditions for the regional wave model. The mesoscale forecast model is run four times a day and provides surface boundary conditions for the sea-surge model and the mesoscale wave model. The global wave model system includes the assimilation of wave height and wind speed observations from the altimeter on ERS-2. The global wave model provides lateral boundary conditions for the regional wave model, which then provides lateral boundary conditions for the mesoscale wave model. The nuclear accident model is run when needed.
7.1 System run schedule
Run |
Model |
Data assimilation |
Hind-cast |
Fore-cast |
Cut-off |
Product available |
Boundary values |
P00 |
Preliminary Global Atmosphere |
2100-0300 |
- |
T+36 |
0155 |
0230 |
- |
W00 |
Regional wave |
- |
12-00 |
T+36 |
0155 |
0240 |
P18, P00 |
M00 |
Mesoscale Atmosphere |
2230-0130 |
|
T+36 |
0200 |
0240 |
P00 |
W00 |
Mesoscale wave |
- |
18-00 |
T+36 |
0200 |
0300 |
P18,P00 |
G00 |
Global Atmosphere |
2100-0300 |
- |
T+120 |
0305 |
0405 |
- |
W00 |
Global wave |
1200-0000 |
12-00 |
T+120 |
0305 |
0420 |
G18,G00 |
S00 |
Stratospheric atmosphere |
2100-0300 |
- |
T+6 |
0505 |
- |
- |
O00 |
Global Ocean |
24 hours |
- |
T+144 |
0520 |
0545 |
G00 |
M03 |
Mesoscale Atmosphere |
0130-0430 |
- |
T+3 |
0545 |
- |
P00 |
U00 |
Global Atmosphere |
2100-0300 |
- |
T+6 |
0715 |
- |
- |
P06 |
Preliminary |
0300-0900 |
- |
T+36 |
0755 |
0830 |
- |
M06 |
Mesoscale Atmosphere |
0430-0730 |
- |
T+36 |
0800 |
0840 |
P06 |
W06 |
Mesoscale wave |
- |
00-06 |
T+36 |
0800 |
0900 |
P00,P06 |
M09 |
Mesoscale Atmosphere |
0730-1030 |
|
T+3 |
1215 |
- |
P06 |
S06 |
Stratospheric atmosphere |
0300-0900 |
- |
T+6 |
1220 |
- |
- |
U06 |
Global Atmosphere |
0300-0900 |
- |
T+6 |
1305 |
- |
- |
SST |
SST Analysis |
0000-2359 |
- |
- |
1310 |
- |
- |
P12 |
Preliminary Global Atmosphere |
0900-1500 |
- |
T+36 |
1355 |
0230 |
- |
W12 |
Regional wave |
- |
00-12 |
T+36 |
1355 |
1440 |
P06,P12 |
M12 |
Mesoscale Atmosphere |
1030-1330 |
- |
T+36 |
1400 |
1440 |
P12 |
W12 |
Mesoscale wave |
- |
06-12 |
T+36 |
1400 |
1500 |
P06,P12 |
G12 |
Global Atmosphere |
0900-1500 |
- |
T+120 |
1505 |
1605 |
- |
W12 |
Global wave |
0000-1200 |
00-12 |
T+120 |
1505 |
1620 |
G06,G12 |
M15 |
Mesoscale Atmosphere |
1330-1630 |
- |
T+3 |
1910 |
- |
P12 |
U12 |
Global Atmosphere |
0900-1500 |
- |
T+6 |
1920 |
- |
- |
P18 |
Preliminary |
1500-2100 |
- |
T+36 |
2000 |
2035 |
- |
M18 |
Mesoscale Atmosphere |
1630-1930 |
- |
T+36 |
2005 |
2040 |
P18 |
W18 |
Mesoscale wave |
- |
12-18 |
T+36 |
2005 |
2105 |
P12,P18 |
S12 |
Stratospheric atmosphere |
0900-1500 |
- |
T+48 |
2105 |
2125 |
- |
M21 |
Mesoscale Atmosphere |
1930-2230 |
- |
T+3 |
0010 |
- |
P18 |
S18 |
Stratospheric atmosphere |
1500-2100 |
- |
T+6 |
0020 |
- |
- |
U18 |
Global Atmosphere |
1500-2100 |
- |
T+6 |
0105 |
- |
- |
NB: The global Atmosphere and wave model run out to T+144 for backup purposes only. The preliminary global atmosphere and regional wave models run out to T+48 for backup purposes only.
7.2 Medium range forecasting system (4-10 days): Global model
7.2.1 Data assimilation
Analysed variables |
Velocity potential, stream function, unbalanced pressure and relative humidity. |
Analysis domain |
Global. |
Horizontal grid |
Half model resolution (see 7.2.2) but using an Arakawa C grid. |
Vertical grid |
Same levels as model (see 7.2.2) but using a Charney-Phillips staggering. |
Assimilation method |
3D Variational analysis of increments (Lorenc et al, 2000). Data grouped into 6-hour time windows centred on analysis hour for quality control. |
Assimilation model |
As global forecast model (see 7.2.2). |
Assimilation cycle |
6 hourly. |
Initialisation |
Increments are introduced gradually into the model using an Incremental Analysis Update (Bloom et al, 1996) over 6-hour period (T-3 to T+3). |
7.2.2 Forecast model
Basic equations Hydrostatic primitive equations with approximations accurate on
planetary scales (White & Bromley, 1995). Fourth order accurate
advection.
Independent variables Latitude, longitude, eta, time.
Dependent variables Horizontal wind components, potential temperature, specific humidity, specific cloud water (liquid and frozen), surface pressure, soil temperature, soil moisture content, canopy water content, snow depth, sea-ice temperature, boundary-layer depth, sea-surface roughness.
Diagnostic variables Geopotential, vertical velocity, convective-cloud base, top, amount, and layer-cloud amounts.
Integration domain Global.
Horizontal grid Spherical latitude-longitude with poles at 90ºN and 90ºS. Resolution:
0.56º latitude, and 0.83º longitude. Variables staggered on Arakawa
B-grid.
Vertical grid 30 levels, hybrid co-ordinates (eta = A/po +B);
layer boundaries at 1.0, 0.994, 0.956, 0.905, 0.835, 0.75, 0.70,0.65,
0.60,0.55, 0.50,0.45,0.41, 0.37,0.34, 0.31, 0.29,0.26,0.21,0.19,
0.165, 0.140,0.115,0.090, 0.065, 0.040, 0.020, 0.010, 0.0005;
levels are (assuming surface pressure of 1000 hPa): 997, 975, 930, 880,827,775,725, 675,625,575, 525, 475,430, 390,355,327, 302,
277,252,227, 202, 177,152,127,102, 77, 52, 30, 15, 4.6 hPa.
Integration scheme Split-explicit finite difference. Adjustment uses forward-backward
scheme, second-order accurate in space and time. Advection uses a
two-step Heun scheme with fourth-order accuracy. Adjustment timestep=133.3s; advection timestep =400s; physics timestep =1200s.
Filtering Fourier damping of mass-weighted winds and mass-weighted increments to potential temperature and humidity. Adapts to strength of wind at each latitude.
Horizontal diffusion Linear fourth order with coefficient K = 2.0 x 107 (but linear, second
order on top level with K = 7.0 x 105) for winds, liquid potential
temperature and total water content. No diffusion where co-ordinate
surfaces are too steep (near orography).
Vertical diffusion Second-order diffusion of winds only between 500 & 150 hPa in Tropics (equatorwards of 30º).
Divergence damping Nil.
Orography Grid-box mean, standard deviation and sub-grid-scale gradients (for
gravity wave surface stress) derived from US Navy 10' dataset.
Orographic roughness parameters linearly derived from standard
deviation, and from 1 km data (N.America) and 100m data (Europe).
Surface classification Sea: global SST analysis performed daily.
Sea ice: Analysis using NCEP SSM/I . Partial cover 0.5 to 1, thickness = 2m, Arctic, 1m Antarctic.
Land: geographical specification of vegetation and soil types that
determine surface roughness, albedo, heat capacity, and surface
hydrology; snow amount from modified monthly climatology of Willmott et al. (1985).
Physics parametrizations:
a) Surface and soil Met Office surface exchange scheme (MOSES I), Cox et al, 1999,
which includes:
b) Boundary layer Turbulent fluxes in lowest 5 layers depend on moist local stability and low-cloud cover (Smith, 1990). Implicit integration scheme. Non-local mixing of heat and moisture in unstable conditions. Form drag effects modelled via an effective roughness length calculated from the silhouette area of unresolved orography and standard deviation of orography height within the grid box.
c) Cloud/precipitation Liquid and ice content included. Large-scale precipitation takes into account accretion and coalescence for rain. Frozen cloud starts precipitating as soon as it forms (Smith, 1990). Evaporation of precipitation depends on phase, temperature and rate.
d) Radiation Fully interactive using 6 bands in the long-wave and 4 in solar calculations. Long-wave gaseous transmission adapted from Morcrette et al. (1986). Fractional cloud in all moist layers and convective tower. Cloud emissivity and optical properties depend on phase and water content (Slingo, 1989).
e) Convection Penetrative mass-flux scheme based on a simple cloud model (Gregory and Rowntree, 1990). Initial mass flux depends on buoyancy. Downdraught representation included. Convective momentum transports included. CAPE closure dependence, with adjustment timescale of 1 hour.
f) Gravity-wave drag Surface stress estimated from sub-grid variance of orography and the orography gradient vector; high drag states, flow blocking, and drag due to trapped lee waves are represented. Vertical stress profile for hydrostatic waves is determined by critical saturation stress law similar to Palmer et al., (1986).
7.2.3 Numerical weather prediction products
RSMC Bracknell issues products on the GTS from the global numerical forecast models using several data formats. The character-based format is GRID (FM47-IX Ext.) and the binary format is GRIB Edition 1 (FM92-X Ext.). NWP model fields are interpolated onto regular latitude-longitude grids arranged in adjacent areas to give global coverage in GTS bulletins that do not exceed the GTS size limit. The regular products from the global atmospheric and wave models are on a 2.5º x 2.5º degree resolution. WAFS bulletins from the atmospheric model in the Thinned GRIB format of 140 km (1.25º x 1.25º degrees at the equator) are available over SADIS, but these bulletins, and the rest of the bulletins making up a full forecast product, are also available over high capacity links on the GTS. Graphical products can also be produced as T.4 faxes and Computer Graphic Metafiles (CGMs). Fields from the atmospheric model include geopotential height, temperature, horizontal wind on all standard levels, vertical velocity and relative humidity on some standard levels, mean sea level pressure and precipitation. From the wave model, height, direction and period of total significant wave, swell and wind-sea are available. Forecast times include the analysed data (T+0) and at 6 or 12 hour steps out to T+120. More detailed information is available in the "List of Numerical Weather Prediction Products Available from Bracknell", published by the Met Office.
In the event of a system failure at Bracknell, model runs can be switched to another supercomputer. Backup procedures are started when it becomes apparent that delays to the current run will lead to failure to meet the output schedules. At present this occurs if there is a delay of 20 minutes to the global model output. If the delay is less than 12 hours, all the normal output is generated from the previous run. The previous T+36 becomes the new T+24, for example, and the previous T+12 becomes the new T+0. If the delay is greater than 12 hours, and the previous run was not completed, then a limited number of products based on output from Washington WMC are issued in place of the normal Bracknell output.
A set of Model Output Statistics Products is generated from NWP global model forecast data from the 00Z and 12Z runs out to 6 days ahead. Day-maximum and night-minimum temperature forecasts for 800 stations world-wide and Probability of Precipitation over 6 and 12 hour periods for 300 European stations are produced. The NWP forecast data are sent to a system called FSSSI (Forecasting for Specific Sites: System Implementation) which contains a relational database, and are then run through a set of Kalman Filters to produce the forecasts. Later, the verifying observations are extracted to the database where the Kalman Filters are updated for the new observations. The forecasts are sent to other product generating systems where the data is formatted as an end product.
The ECMWF Ensemble Prediction System (EPS) is utilised for medium range forecasting. Ensembles are also run for monthly and seasonal forecasts (see Sections 7.5 and 7.6).
Output from the EPS is post-processed to provide forecasters with numerous chart displays including spaghetti diagrams, ensemble means, individual ensemble members and clusters of members. Charts showing extra-tropical cyclone tracks in ensemble members are also available. Probability forecast information for 41 synoptic sites around the British Isles are also generated, along with a comprehensive verification system for their assessment.
A new system scans the ensemble for probabilities of severe weather (severe gales, heavy rain or snow) and issues automatic alerts to forecasters when defined probability thresholds are exceeded. This system incorporates considerable calibration and also sophisticated calculation of probabilities over regions and time windows, to assess probabilities explicitly in the form required to support the UK National Severe Weather Warning Service. The system is currently under operational trial and is expected to become fully operational in 2001.
7.3 Short-range forecasting system (0-72 hrs): Mesoscale model
7.3.1 Data assimilation
The data assimilation scheme for the mesoscale model is similar to that for the global model except in the following:
Analysis variables As global (see 7.2.1) but also includes aerosol content.
Analysis domain |
As model integration domain (see 7.3.2). |
Horizontal grid Vertical grid Assimilation method |
Half model resolution (see 7.3.2), but using an Arakawa C grid. 3D Variational analysis of increments for 'conventional' data (Lorenc et al, 2000), with nudging for cloud and rainfall data. Data grouped into 3-hour time windows centred on analysis hour for quality control. |
Assimilation model |
As mesoscale forecast model (see 7.3.2) |
Assimilation cycle |
Continuous sequence of 3-hourly mesoscale assimilation cycles. Increments from 'conventional' data are introduced gradually into the model using an Incremental Analysis Update (Bloom et al, 1996) over a 2-hour period (T-1 to T+1), while increments from cloud and rainfall data are added by nudging. |
Data |
Screen temperature, humidity, visibility and surface wind data are assimilated by the mesoscale model. A 3-dimensional 'MOPS' cloud fraction analysis, derived from satellite imagery and surface reports, is assimilated (Macpherson et al., 1996). An hourly precipitation rate analysis, derived from radar, is assimilated by latent heat nudging (Jones and Macpherson, 1997). The precipitation analysis is also used to calculate assimilation increments to the canopy water, soil moisture and snow depth fields. A weekly analysis of soil moisture content is performed from 'data' produced by the 'MORECS' agricultural model for the UK. |
7.3.2 Forecast model
The mesoscale forecast model is identical to the global model in all respects except the following:
Integration domain British Isles and surrounding sea areas, near continental Europe and southern Norway (approximately 64ºN-44ºN, 12ºW-13ºE).
Horizontal grid Spherical rotated latitude-longitude with pole at 37.5ºN, 177.5ºE. Resolution: 0.11º.
Vertical grid 38 levels, hybrid co-ordinates (eta = A/p0 +B);
layer boundaries at 1.000, 0.9976, 0.9929, 0.9835, 0.9719, 0.958, 0.940, 0.921, 0.901, 0.880, 0.858, 0.835, 0.810, 0.780, 0.745, 0.705,0.660, 0.610, 0.555, 0.500, 0.450, 0.410,0.370, 0.340,0.310, 0.290, 0.265, 0.240,0.215,0.190, 0.165, 0.140, 0.115, 0.090,0.065, 0.040, 0.020, 0.010, 0.0005;
levels are (assuming surface pressure of 1000 hPa): 999, 995, 988, 978, 965, 949, 930, 911, 890, 870, 846, 822, 795, 762, 725, 682, 635, 582, 527, 475, 430, 390, 355, 327, 302, 277, 252, 227, 202, 177, 152, 102, 77, 52, 30, 15, 4.6 hPa.
Timestep Adjustment timestep = 25s; advection timestep = 75s; physics timestep = 300s.
Horizontal diffusion Linear fourth-order with coefficient K = 1.9 x 106 for winds, liquid water, potential temperature, and total water content. No diffusion where co-ordinate surfaces are too steep (near orography).
Vertical diffusion None.
Orography Grid-box mean and variance derived from 5' NCAR dataset. Orographic roughness parameters derived from 100m data.
Boundary values Specified from preliminary global forecast model with same data time (forecasts from 00, 06, 12 and 18 UTC)
Physics parametrizations:
a) Surface Met Office surface exchange scheme (MOSES II), Cox et al, 1999, which includes:
b) Boundary layer A new turbulent mixing scheme (Lock et al, 1999; Martin et al, 1999). Includes representation of non-local mixing driven by both surface fluxes and cloud top processes in unstable layers, either coupled to or decoupled from the surface; also includes an explicit entrainment parametrization. A moist conserved variable formulation is used - suitable for both dry and cloud layers.
c) Cloud/precipitation Cloud ice is treated prognostically, with 11 transfer terms between cloud ice, liquid and precipitation products. (Wilson and Ballard, 1999)
d) Radiation Edwards-Slingo flexible two-stream code (1996). Calculated on chequerboard pattern for every other grid box and heating rates interpolated from same land-sea types. Updated hourly with solar angle updated each timestep.
e) Convection Updated version of Gregory and Rowntree scheme (1990) to include downdraught parametrization and revised evaporation formulae dependent on the precipitation rate.
f) Gravity-wave drag None.
7.3.3 Numerical weather prediction products
As described in section 7.2.3, except for the following:
NWP model fields are interpolated onto regular latitude-longitude grids arranged in adjacent areas covering Europe and the North Atlantic. The regional products from the atmospheric model are currently on either a 1.25º x 1.25º or a 2.5º x 2.5º degree resolution. The regional wave model products are on a 1.25º x 1.25º degree resolution. Forecast times include the analysed data (T+0) and at 3 or 6 hour steps out to T+36. Backup procedures are started when there is a delay of 10 minutes to the regional output. If the delay is less than 6 hours, all the normal output is generated from the previous run. The previous T+36 becomes the new T+30, for example, and the previous T+12 becomes the new T+6. If the delay is greater than 12 hours, and the previous run was not completed, then products from the run 12 hours before are used.
7.3.4 Operational techniques for application of NWP products
7.4 Specialised forecasts
Nowcasting system
Nimrod produces analyses and forecasts of precipitation and supplementary weather parameters (including precipitation type, visibility, snow probability and lightning rate), at 5 km resolution, for the period T+0 to T+6 hours. Forecasts are produced using a combination of linear extrapolation and model wind advection, with precipitation forecasts from the mesoscale model used to introduce an element of growth/decay. In addition, analyses and forecasts of cloud amount (3-D), cloud base and cloud top height, wind speed and direction are generated. These products are generated hourly at a resolution of 15 km. The Nimrod cloud and precipitation analyses are used as inputs to the mesoscale model assimilation scheme.
Grid UK national grid: 2 and 5 km resolution for precipitation products; 15 km resolution for cloud products. The domain is an approximation to the mesoscale model domain (roughly 44ºN-64ºN, 12ºW-13ºE).
Data inputs Radar imagery (from the network of 15 sites within the UK, Ireland and Jersey), Meteosat visible and infrared imagery, mesoscale model forecast fields and surface weather reports.
Forecast timestep 10 and 15 minutes for precipitation forecasts; 60 minutes for cloud and visibility forecasts.
Special features Radar rain rates automatically corrected for the effects of bright-band, range and orographic growth using a physically based method (Kitchen et al., 1994)
Stratospheric Model
A stratospheric data assimilation system has been run at the Met Office for a number of years. A research data assimilation system was started in October 1991 based on the Met Office Analysis Correction (AC) scheme. This research system was the basis for the operational stratospheric data assimilation system which was started in November 1994. In November 2000, major changes were made to the operational stratospheric suite: the AC scheme was replaced by a 3-D variational scheme; satellite radiances were assimilated instead of retrieved temperature profiles; and a daily 48-hour forecast was introduced.
Model type |
Low horizontal resolution version of the standard global forecast model (section 7.2), but with additional stratospheric levels |
Integration domain |
Global |
Levels |
40 hybrid co-ordinate levels |
Grid |
Horizontal resolution: 2.5 degrees latitude by 3.75 degrees longitude |
Data Assimilation |
3-D variational data assimilation scheme (Lorenc et al, 2000). |
Stratospheric Model
A stratospheric data assimilation system has been run at the Met Office for a number of years. A research data assimilation system was started in October 1991 based on the Met Office Analysis Correction (AC) scheme. This research system was the basis for the operational stratospheric data assimilation system which was started in November 1994. In November 2000, major changes were made to the operational stratospheric suite: the AC scheme was replaced by a 3-D variational scheme; satellite radiances were assimilated instead of retrieved temperature profiles; and a daily 48-hour forecast was introduced.
Tropical cyclone forecasts
Initialisation of TC's is achieved by the creation of bogus data, which are fed into the numerical forecast model. TC advisory bulletins received on the GTS from various TC warning centres are used to provide the input data to this process. The creation of TC bogus data is totally automated, but forecasters in the National Meteorological Centre (NMC) at the Met Office have the facility to over-ride the automatic system and create their own bogus data if required. Full details of the bogus technique may be found in Heming et al, 1995.
Transport and dispersion model
A model for medium and long-range transport and dispersion (NAME, version 4.7) is available to be run in the event of a major atmospheric release of hazardous pollutants, such as in a nuclear emergency or volcanic eruption. It can also be used for routine air pollution problems, such as episode studies and forecasting air quality. The model provides forecasts of concentrations in the boundary layer and at upper levels, as well as wet and dry deposition to the surface. It uses analysis and forecast fields from the global and mesoscale atmospheric models maintained in on-line archives. The NAME model may be run at any time in hindcast or forecast mode.
Model type Three-dimensional Lagrangian multiple-particle Monte Carlo model simulating the medium or long-range transport, dispersion and deposition of airborne pollutants.
Integration domain Global or UK mesoscale, nested as required.
Model grid Identical to the global and UK mesoscale models, but with some vertical levels omitted. The transport model can access fields from both input models simultaneously, with an option to use the best resolution available at every particle position. The output grid can be customer-defined, and of any resolution.
Dynamical input Meteorological fields from the global or UK mesoscale models.
Integration scheme Forward timestep, determined by the diffusion scheme near to the source, but with an option for definition by the user in the longer ranges.
Parametrization Comprehensive near-source random walk diffusion scheme with option of uniform or non-homogeneous vertical profiles of the turbulent velocity variances and, for the convective boundary layer, an option for non-Gaussian statistics. Simpler, homogeneous scheme at longer ranges, or an option for basic K-diffusion. There are also schemes for low-frequency wind meandering, plume rise, gravitational settling, the venting of pollutants from the boundary layer by strong convection, and a novel technique allowing for small-scale entrainments at the boundary layer top. Radioactive decay; wet and dry deposition, with provision for conversions from gas to particulate; turbulent (occult) deposition over hills. Sulphur and nitrogen chemistry, including aqueous phase.
Special features Utilises high-resolution (6 km) rainfall rates, derived from radar products, for detailed wet deposition over NW Europe. Source attribution scheme for identifying origin of material predicted at a given receptor.
Global ocean model - FOAM (Forecast Ocean Atmosphere Model)
Model type |
Developed from Bryan-Cox ‘level’ model on Arakawa B-grid. Includes a Kraus-Turner mixed layer scheme, and a thermodynamic/simple advection sea ice model. |
Integration domain |
Global. |
Horizontal grid |
1º x 1º. |
Vertical grid |
20 levels. 10 of the levels are in the top 300m, the deepest is at 5192m. |
Data assimilation |
Based on the Met Office analysis correction scheme. Assimilates temperature profile data, and sea surface temperature data (in-situ and AVHRR). Gridded SSMI sea-ice concentration data are assimilated, using a nudging technique. |
Surface fluxes |
From global NWP model, 6-hourly. |
Wave hindcast and forecasting system: Global wave model
Model type |
Coupled discrete (SWAMP, 1985) |
Integration domain |
Global. |
Grid |
Spherical latitude-longitude from 80.2778ºN to 79.166ºS. Resolution: 5/9º latitude, 5/6º longitude. |
Frequency resolution |
13 frequency components spaced logarithmically between 0.04Hz and 0.324Hz. |
Direction resolution |
16 equally spaced direction components. |
Data assimilation |
ERS-2 altimeter wave-height observations are assimilated into the global wave model, using the altimeter wind-speed to separate wind-sea and swell. The assimilation scheme (Thomas, 1988, and Stratton et al., 1990) is a variant of the analysis-correction scheme of Lorenc et al. (1991). After assimilation the model wave height matches the analysed wave height, the model wind-sea matches the analysed wind-speed, and the pattern of the spectrum remains similar to that before assimilation. |
Integration scheme |
Modified Lax-Wendroff. Source terms timestep = 1800s; advection timestep frequency dependent. |
Boundary forcing |
Winds at 10m, updated hourly |
Surface classification |
Sea ice analyses as in the global atmospheric model. |
Physics parametrizations: |
Linear growth (Phillips, 1958); exponential growth (Snyder, 1981); white-capping dissipation (Komen et al., 1984). Non-linear transfer of wave energy is parametrized by enforcing a JONSWAP spectral shape on the wind-sea. A parametrization of directional relaxation in turning winds is included, and a term accounting for the great-circle turning of swell energy is applied. For wind speeds lower than 7.3 ms-1, a parametric growth term is used to calculate windsea growth. For all but actively growing windsea, the dissipation coefficient is reduced by one half of the specified value. Shallow water terms are included (shoaling, bottom friction, refraction) |
Wave hindcast and forecasting system: Regional wave model
Apart from having no data assimilation, the formulation of the regional wave model is identical to the global wave model in all respects except the following:
Model type |
Coupled discrete; depth dependency specified to 200m with 2m resolution. |
Integration domain |
European continental shelf and Mediterranean, Baltic and Black Seas. |
Grid |
Spherical latitude-longitude from 67.7ºN to 30.5ºN, and from 14.1ºW to 41.9ºE. Resolution: 0.25º latitude, 0.4º longitude. |
Source terms timestep |
1800s. |
Boundary forcing |
1. Winds at 10m. Updated hourly. 2. Spectral values at lateral boundaries from global wave model. Updated hourly. |
Surface classification |
No sea ice. |
Physics parametrizations: |
Identical to global model, without great-circle turning of swell. |
Wave hindcast and forecasting system: Local wave model
A new UK Waters wave model has been implemented. The wave model uses the same physics as the regional wave model, with the addition of time-varying wave-current interactions, taking surface currents from the operational storm surge model. The model is set up at 1/9 by 1/6 degree resolution covering 48-63N, 12W-13E, the same grid as the operational storm surge model. The model is run four times daily, for a 48 hour forecast under mesoscale 10m winds. A separate 5-day forecast, without currents, is run twice daily using global NWP winds.
Model type Coupled discrete; depth dependency specified to 200m with 2m resolution.
Integration domain NW European continental shelf.
Grid Spherical latitude-longitude from 48N to 63N, 12W to 13E Resolution: 1/9º latitude, 1/6º longitude.
Boundary forcing 1. 10m winds from mesoscale NWP, updated hourly (for 48 hour forecast four times daily). Winds from global NWP for 120 hour forecast twice daily.
2. Spectral values at lateral boundaries from global wave model. Updated hourly.
Surface classification No sea ice. Surface currents from operational storm surge model (hourly) (not used in 5-day forecast)
Physics
parametrizations Identical to global model, without great-circle turning of swell. Plus calculation of effect of time-varying currents on wave energy spectrum.
Shelf seas forecast model
The Proudman Oceanographic Laboratory (POL) baroclinic shelf seas model (Proctor and James, 1996) covering the NW European shelf area, at 1/9 by 1/6 degree resolution between 48-63N, 12W-13E, and with 15 sigma levels in the vertical, is run using surface forcing from the mesoscale numerical weather prediction model (3-hourly average for surface heating, hourly winds and pressures). Climatological values of temperature and salinity are applied at the deep ocean boundary. Tides are forced by specifying the boundary elevation using 15 harmonic constituents. The model runs once daily, for a 24 hour hindcast followed by a 48 hour forecast. There is no data assimilation.
Model type Baroclinic, piecewise parabolic advection scheme, bulk Richardson number vertical mixing. Sigma co-ordinate
Integration domain NW European continental shelf.
Grid Spherical latitude-longitude from 48ºN to 63ºN, and from 12ºW to
13ºE. Resolution: 1/9º latitude, 1/6º longitude.
Boundary forcing 1. Hourly winds and pressures, 3-hourly averaged heat flux, from mesoscale NWP.
2. Deep ocean temperature and salinity profiles from monthly climatology.
3. River inflows. Annual mean volume flow rate.
4. Tidal elevations from 15 constituents
Surface classification No sea ice. No wetting or drying.
Storm Surge model
A depth-averaged storm surge model, developed by the Proudman Oceanographic Laboratory, is run operationally on behalf of MAFF (Ministry of Agriculture, Fisheries and Food) for the Storm Tide Forecasting Service. The model is implemented on a grid at 1/9 by 1/6 degree resolution covering 48-63N, 12W-13E, and is forced at the deep ocean boundaries by 15 tidal harmonic constituents. The model is run 4 times daily, using hourly values of surface pressure and 10m winds from the mesoscale NWP model to provide a 36 hour forecast.
7.5 Extended range forecasts (10 days to 30 days)
Extended range and experimental seasonal range forecasts (Section 7.6) are produced from the same 4-month-range, 9-member AGCM ensemble integrations forced with persisted Sea Surface Temperature (SST) anomalies. Forecasts are produced weekly on Thursdays.
Model: The HadAM3 climate version of the Met Office’s Unified Model (UM Vn4.5) is used (Pope et al. 2000). The resolution used is 2.5o latitude, 3.75o longitude and 19 vertical levels. The timestep is 30 minutes. The model is run in a 9-member ensemble.
Atmospheric initial conditions: Initial conditions for the ensemble are provided by consecutive operational NWP analyses at 6-hour intervals. The first member being initialised with the 00Z analysis each Tuesday and the final member with the 00Z analysis on the following Thursday.
SST and sea-ice forcing: SST anomalies calculated from the Reynolds SST analysis for the 4-week period lagging the initialisation date by 10 days are persisted throughout the integration, updating every 24hrs. SST forcing is the same for all members. Projected changes in sea-ice cover are also represented.
Treatment of land surface variables: Initial conditions for soil moisture, soil temperature and snow cover are taken from climatology. Land surface exchanges are represented using the MOSES scheme (Cox et al. 1999).
Forecast variables: The main forecast variables are mean, maximum and minimum temperature, accumulated precipitation and sunshine amount averaged over three forecast periods; days 4-10, days 11-17 and days 18-31. For each ensemble member, global forecast values are derived from direct averaging of daily model output. For the UK region only, values are also derived using regression equations on the forecast period-averaged PMSL field and observed local SST.
Model calibration: Forecast anomalies are expressed relative to a model climatology defined for each month of the year from a set of integrations initialised at the beginning of each month over the 15-year period 1979-1993.
Forecast formats: Temperature and rainfall forecasts are mainly presented in terms of equi-probable quintile categories; Well Below, Below, Near Normal, Above, Well Above. Tercile categories are used for some forecasts. The forecast is expressed both in terms of the probability of each category and a single deterministic forecast based on the ensemble mean.
7.6 Long range forecasts (30 days up to 2 years)
The model ensemble system used for long (seasonal) range forecasts is identical to that used for extended range forecasts (Section 7.5). The seasonal forecast products are experimental and are available to National Met. Services through a password protected internet site.
Forecast variables: Forecasts are provided for anomalies in 3-month-average 850 hPa temperature (as a proxy for surface temperature) and precipitation. Forecasts at zero lead (months 1-3 of the integration) and 1 month lead (months 2-4 of the integration) are produced.
Model calibration: Forecast anomalies are expressed relative to a model climatology defined for each month of the year from a set of 9-member ensemble integrations initialised at the beginning of each season over the 19-year period 1979-1997. The same set of integrations has been analysed to assess seasonal prediction skill and to generate "skill templates" (see below).
Forecast format: Both probability and deterministic forecasts are produced. For probability forecasts a two category format is used, i.e. probability that the anomaly will be above or below zero (based on the ensemble distribution). For deterministic forecasts the anomaly sign and magnitude is provided (based on the ensemble mean). Products are provided in map format for the globe and a number of regional areas and with optional skill templates, which mask out regions in which the model currently has no significant skill.
8. Verification of prognostic products
Statistic |
Parameter |
Area |
Verified against |
T+24 |
T+72 |
T+120 |
RMS error(m) |
Z 500 |
N.Hem |
Analyses |
12.35 |
32.09 |
57.56 |
RMS error(m) |
Z 500 |
S.Hem |
Analyses |
17.21 |
43.85 |
73.91 |
RMS error(m) |
Z 500 |
N.America |
Observations |
14.49 |
33.63 |
60.75 |
RMS error(m) |
Z 500 |
Europe |
Observations |
14.81 |
32.18 |
61.17 |
RMS error(m) |
Z 500 |
Asia |
Observations |
15.95 |
28.07 |
43.93 |
RMS error(m) |
Z 500 |
Aus/NZ |
Observations |
14.15 |
25.43 |
43.12 |
RMSVW error (m/s) |
W 250 |
N.Hem |
Analyses |
4.92 |
10.19 |
15.58 |
RMSVW error (m/s) |
W 250 |
S.Hem |
Analyses |
5.76 |
11.97 |
17.72 |
RMSVW error (m/s) |
W 250 |
N.America |
Observations |
6.78 |
12.05 |
18.26 |
RMSVW error (m/s) |
W 250 |
Europe |
Observations |
6.27 |
11.16 |
17.77 |
RMSVW error (m/s) |
W 250 |
Asia |
Observations |
6.69 |
10.51 |
14.11 |
RMSVW error (m/s) |
W 250 |
Aus/NZ |
Observations |
6.75 |
10.76 |
15.33 |
RMSVW error (m/s) |
W 850 |
Tropics |
Analyses |
2.30 |
3.44 |
4.10 |
RMSVW error (m/s) |
W 250 |
Tropics |
Analyses |
4.21 |
6.75 |
8.27 |
RMSVW error (m/s) |
W 850 |
Tropics |
Observations |
4.13 |
4.95 |
5.40 |
RMSVW error (m/s) |
W 250 |
Tropics |
Observations |
5.98 |
7.44 |
8.63 |
9. Plans for the future
9.1 Computer systems
New applications will be developed to meet new user requirements. Over the coming year or so we will also be exploring how best to optimise the use of Horace and the Met Office’s Windows NT visualisation and production platform, Nimbus, which is used primarily in the local weather centres and military stations around the UK. We will also be investigating the possibility of porting Horace to run on the Linux operating system.
9.2 Data assimilation.
An upgrade to the current global data assimilation system is anticipated in the early spring. Later in the year a major upgrade of the system is anticipated in order to support the new version of the modelling system. Some possible candidates for inclusion in the spring upgrade are:
Upgrades to the mesoscale data assimilation system during 2001 may include:
9.3 Atmospheric forecast models
9.4 Nowcasting System
The resolution of some of the rainfall products will be increased to 1km. Data from more European radars will be included in the rainfall analysis. An improved gauge adjustment scheme for the radar data will introduced. A sophisticated soil moisture scheme will be incorporated into Nimrod using Nimrod and model forecasts as input. An improved version of the idealised model of convection used in the GANDOLF convective forecasting system will be introduced and the rainfall forecasts from Gandolf and Nimrod will be fully integrated.
9.5 Tropical cyclone forecasting
The sources of incoming advisories for TC initialisation procedures will continue to be evaluated and changes made to ensure the use of the best input data. Tests of enhancements to the TC initialisation procedure itself are ongoing and changes will be implemented if a positive impact on TC forecasts is established.
9.6 Transport and dispersion model
9.7 Ocean, wave and surge forecast models
Wave models
Wave energy spectra from the ERS-2 SAR UWA observations are routinely retrieved, with global coverage, using the iterative retrieval scheme developed by Hasselman et al, at the Max Planck Institute for Meteorology, Hamburg.
Ocean Forecast models.
A 1/3 degree model of the Atlantic and arctic is being developed, with a southern boundary at 30S nested into global FOAM. Models at 1/9 degree nested into the 1/3 degree Atlantic model are also under development. Methods to assimilate altimeter sea surface height data are being developed and applied. The 1/3 degree Atlantic model is being prepared for operational implementation during second quarter 2001
Shelf Seas forecast models
The shelf seas model is being upgraded to use a hybrid vertical co-ordinate, to improve representation of the surface layers in deeper water. The new model will use a Mellor Yamada turbulent mixing scheme, in place of the present bulk Richardson number scheme, and will be applied over a wider area (40N to 60N, 20W to 13E) The revised model will use the POLCOMS formulation and code.
10. References
Bloom, S.C. , Takaka, L.L., Da Silva, A.M. and Ledvina, D., 1996, Data assimilation using incremental analysis updates. Mon. Wea. Rev., 124, 1256-1271
Cox, P.M., R.A. Betts, C.B. Bunton, R.L.H. Essery, P.R. Rowntree and J. Smith, 1999: The impact of new land surface physics on the GCM simulation of climate and climate sensitivity. Climate Dynamics 15, 183-203.
Edwards, J.M. and Slingo A., 1996:Studies with a flexible new radiation code. Part I. Choosing a configuration for a large-scale model. Quart. J. Roy. Meteor. Soc., 122 pp.689-719
English, S.J. R. J. Renshaw, P. C. Dibben, A. J. Smith, P. J. Rayer, C. Poulsen, F. W. Saunders and J. R. Eyre, 2000: A comparison of the impact of TOVS and ATOVS satellite sounding data on the accuracy of numerical weather forecasts. Quart. J.R. Meteorol. Soc. Vol. 126 No. 569 pp2911
Gregory, D. and P.R. Rowntree, 1990: A mass-flux convection scheme with representation of cloud ensemble characteristics and stability-dependent closure. Mon. Wea. Rev., 118, pp1483-1506.
Heming, J.T.,J.C.L.Chan & A.M.Radford (June 1995). A new scheme for the initialisation of tropical cyclones in the UK Meteorological Office global model. Meteorological Applications, 2, 171-184.
Jones,C.D and Macpherson,B., 1997: A Latent Heat Nudging scheme for the Assimilation of Precipitation Data into an operational Mesoscale Model. Meteorol. Apps, 4, 269-277
Kitchen, M., R. Brown and A. G. Davies, 1994: Real-time correction of weather radar data for the effects of bright band, range and orographic growth in widespread precipitation. Quart. J. Roy. Meteor. Soc., 120, 1231-1254
Komen, G., K. Hasselmann, and S. Hasselmann, 1984: On the existence of a fully developed windsea spectrum. J. Phys. Oceanogr., 14, pp1272-1285.
Lock, A.P., A.R. Brown, M. R. Bush, G. M. Martin and R.N.B. Smith, 1999: A new boundary layer Mixing scheme. Part I: Scheme description and single column tests. Submitted to Mon. Wea. Rev.
Lorenc, A.C., and O.Hammon, 1988: Objective quality control of observations using Bayesian methods. Theory
and a practical implementation. Quart. J. R. Meteorol. Soc., 114, pp515-543.
Lorenc, A.C., Ballard, S.P., Bell, R.S., Ingleby, N.B., Andrews, P.L.F., Barker, D.M., Bray, J.R., Clayton, A.M., Dalby, T., Li, D., Payne, T.J. and Saunders, F.W. 2000. The Met Office Global 3-Dimensional Variational Data Assimilation. Quart. J.R. Meteorol. Soc. Vol. 126 No. 570 pp2991
Macpherson, B., Wright,B.J., Hand,W.H. and Maycock,A.J., 1996: The impact of MOPS moisture data in the UK Meteorological Office Mesoscale Data Assimilation Scheme Mon. Wea. Rev., 124,1746-1766
Martin G. M., M. R. Bush, A.R. Brown, A. P. Lock, and R.N.B. Smith, 1999: A new boundary layer Mixing scheme. Part II: Tests in climate and mesoscale models. Submitted to Mon. Wea. Rev.
Morcrette, J.-J., L.D.Smith, and Y.Fouquart, 1986: Pressure and temperature dependence of the adsorption in long-wave radiation parameterisations. Beitr. Phys. Atmosph., 59, pp455-469.
Palmer, T.N., G.J. Shutts, and R. Swinbank, 1986: Alleviation of a systematic westerly bias in general circulation and numerical weather prediction models through an orographic gravity wave drag parametrization. Q. J. Roy. Meteorol. Soc., 112, pp1001-1039.
Phillips, O.M., 1958: The equilibrium range in the spectrum of wind generated waves. J. Fluid Mech., 4, pp426-434.
Pope, V.D., M.L. Gallani, P.R. Rowntree and R.A. Stratton, 2000: The impact of new physical parametrizations in the Hadley Centre climate model: HadAM3. Climate Dynamics, 16,123-146.
Proctor, R. and I. D. James (1996) A fine resolution model of the Southern North Sea. Journal of Marine Systems 8 (1996) 285-295
Slingo, A., 1989: A GCM parametrization for the short-wave radiative properties of water clouds. J. Atmos. Sci., 46, pp1419-1427.
Smith, R.N.B., 1990: A scheme for predicting layer clouds and their water contents in a general circulation model. Q. J. Roy. Meteorol. Soc., 116, pp435-460.
Snyder, R.L., F.W. Dobson, J.A. Elliot, and R.B. Long, 1981: Array measurements of atmospheric pressure fluctuations above surface gravity waves. J. Fluid Mech., 102, pp1-60.
Stratton, R.A., D.L.Harrison, and R.A.Bromley, 1990: The assimilation of altimeter observations into a global wave model. AMS 5th Conference on Satellite Meteorology and Oceanography, London, 1990, pp108-109.
SWAMP (Sea Wave Modelling Project), 1985: An intercomparison study of wind wave prediction models. Part I: Principal results and conclusions. Ocean Wave Modelling. Plenum Press, 256pp.
Thomas, J.P., 1988: Retrieval of energy spectra from measured data for assimilation into a wave model. Q.J. Roy. Meteorol. Soc., 114, pp781-800.
White, A.A and R.A.Bromley, 1995: Dynamically consistent, quasi-hydrostatic equations for global models with a complete representation of the Coriolis force. Q.J. Roy. Meteorol. Soc., 121, pp399-418.
Willmott, C.J., C.M. Rowe and Y. Mintz, 1985: Climatology of the terrestrial seasonal water cycle. J. Clim., 5, pp589-606.
Wilson, D. R. , and S. P. Ballard, 1999:A microphysically based precipitation scheme for the UK Meteorological Office Unified Model. Q.J. Roy. Meteorol. Soc., 125, pp.1607-1636.