© Crown copyright Met Office Weather and climate modelling workshop, INPE, Cachoeira Paulista Met...

© Crown copyright Met Office

Weather and climate modelling workshop, INPE, Cachoeira Paulista

Met Office Unified ModelGeorge Pankiewicz, December 2008


Contents

• The need for a Unified Model

• UM formulation & NWP-related plans

• Climate models, coupling & seamless prediction

• UM licensing and collaboration

• The ACCESS modelling system


The need for a Unified Model


The same model formulation is used for all models from climate scale to mesoscale

The Unified Model

Climate modelling: input into IPCC reports(Coupled Atmosphere-Ocean models)

Seasonal forecasting:For commercial and business customers

NWP:Public Weather ServiceWAFC, Commercial ……


UM overview

• A suite of earth system modelling software: UM, VAR, OPS, FCM, SCS …

• Supports global and regional NWP and climate prediction

• In operational use at the Met Office since 1992

• Non-hydrostatic formulation since 2002

• Fortran 90 standards; suitable for Massively Parallel Processing

• R&D funding by the UK Public Weather Service Customer Group is ~£15M per year


Met Office modelling 1988

Climate modelling

• 11L 250km

• “sophisticated” physics

NWP

• coarse mesh global model 15L 150km

• fine mesh NAE LAM 15L 75km

• “simpler” physics, different dynamical integration

• 15km mesoscale model more different dynamics & physics!

New requirements e.g. ocean-atmosphere coupling


The need for a Unified Model

The problem:

• Control, file & output structures different & machine specific

• Separate scientific formulation and development teams

The requirements

• Climate model data processing during a run (hence Stash)

• More resources to develop model physics in NWP teams

• Use of programming and change control standards

• A software application system (GUI)

• To avoid duplication of effort - 1 platform, 1 model

… so the Unified Model project commenced in July 1989


UMUI input panel


UM development 1991-2008

• UM1.0 Oct 1991 - main components ready

• UM2.0 - 2.7: error & instability corrections, HadAM1

• UM3.0 - 3.4: First PUM, GRIB, notable speedup, HadAM2a

• UM3.5 - 4.0: UMUI, Edwards-Slingo, HadAM2b

• UM4.1 - 4.4: MOSES, MPP capability

• UM4.5: SCM, CO2 coupling (often used by researchers)

• UM5.0 - 5.5: NH semi-Lagrangian dynamics, F90, early PC2

• UM6.0 - 6.2: PC2 improvements, makebc, fieldcalc

• UM6.3: FCM introduced, physics for HadGEM1

• UM6.4 - 6.6: 70L improvements, FLUME, UKCA


UM formulation & NWP-related plans


NAE12Km 38 L4DVAR60 hour forecast 4 times per day24member EPS at 24km

UK44km 70 L3DVAR36 hour forecast 4 times per day

Operational NWP Models: 2008Global

40Km 50 L4DVAR60 hour forecast twice/day144 hour forecast twice/day+EPS 24member at 90km


MOGREPS: Regional EPS

45kt winds force 90+ boats to retire from Fastnet race


Current Observation Usage

AircraftSonde +

SatwindSurface



ATOVS Scatwind

SSMISSMIS



Ground GPSGPS RO

IASIAIRS


Data Volumes Assimilated: 6 hour window 00 UTC 20th October 2008

Tamps 631 Scatwinds: Seawinds 5431PILOTs 305 Scatwinds: ERS 590Wind Profiler 526 Scatwinds: ASCAT 8996Land Synops 16415 SSMI/S 3309Ships 2551 SSMI 4344

Buoys 5471 ATOVS:METOP 12042

Amdars 14701 ATOVS: NOAA18 10949Aireps 882 ATOVS: NOAA17 3942Satwinds: JMA (18th Oct) 1021 ATOVS: NOAA15 9520Satwinds: NESDIS 3979 AIRS 4233Satwinds: EUMETSAT 2771 IASI 2777TC Bogus 132 GPS-RO 770

Ground: GPS-RO (in NAE) 1397


The Operational Suite: Simplified Schedule of Atmosphere Models. Excluding Update Cycles

00:00

02:00

04:00

06:00

08:00

10:00

12:00

NAE 00z+01:30; T+60

NAE 06z+01:30; T+60

GL 00z+02:45; T+144

GL 06z+02:45; T+66

UK 03z+01:20; T+36

UK 09z+01:20; T+36

LBC

LBC

LBC

LBC

LBC


Complete List of Suites (Dec 2008)

UM - Atmosphere UM - Ocean Non-UM

Global 40km

NAE 12km

UK 4km

Sub-UK 1.5km

Southern Asia 16km

Falklands 12km

Africa 20km

German 4km

Global EPS (24*90km)

NAE EPS (24*24km)

NEMO FOAM at 1/4° Global

1/12° North Atlantic

1/12° Mediterranean

1/12° Indian Ocean

NW Europe Shelf-Seas

MRCS Shelf-Seas

Irish Sea Shelf-Seas

Persian Gulf Shelf-Seas

Global WaveWatch III

NAE WaveWatch III

Persian Gulf WaveWatch III

CS3 Shelf Surge

Nested Surge models (Bristol Channel, Severn River and South Coast)

OSTIAThe above suites might run as often as 8 times per day (e.g. Global Atmosphere Suite which runs 6 hourly but with a repeated data assimilation cycle to capture late data)


Data Assimilation Features GL NAE UK4

4DVAR Rawlins et al, 2007

•Perturbation forecast (PF) and adjoint models to provide explicit representation of the time dimension within a 6-hour data window.

•PF model has simplified linearised physics, including a very simple boundary layer, cloud latent heat release, convection and large-scale precipitation

•Weak constraint digital filter to the perturbation forecast (Gauthier and Thépaut, 2001), penalising high frequencies using a pressure-based energy norm.

Horizontal Grid 1/3 of Model Grid

3DVAR Lorenc et al 2000

•3 hour cycle

•Iinitialisation by using the Incremental Analysis Update scheme (Bloom et al., 1996)

VISIBILITY (Clark et al., 2008): Variational assimilation of visibility observations

LAM boundaries: constraint of zero analysis increments is specified. This allows a double sine transform to be used for the horizontal filtering of the control variables.

MOPS: (Macpherson, 1996)

RH Nudging to improve fit to cloud; latent heat nudging using precipitation analysis


Surface Analysis FeaturesGL NAE UK4

SOIL MOISTURE (Best and Maisey 2002),

The global soil moisture is updated using a nudging scheme based on errors in NWP forecasts of screen level temperature and humidity.

(Away

from UK)

SOIL MOISTURE (Smith et al. 2004)

For UK, we rely on the soil moisture estimate from the NIMROD nowcasting scheme This uses observations to estimate the fluxes driving an off-line version of the Met Office land surface and surface exchange scheme (MOSES).

SSTSea-surface temperatures are taken from the Met Office’s Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA) system.


Model Features Comparison

GL (25km) GL (40km) NAE UK4 UK1.5

Horizontal/Vertical advection

cubic cubic cubic cubic cubic

Vertical advection (q) quintic quintic cubic cubic cubic

q horizontal diffusion target target target 4th order 4th order

u, v, theta horizontal diffusion

4th order 4th order None 4th order 4th order

u, v vertical diffusion Selected levels/latitude

s

Selected levels/latitu

des

No No No

Model timestep 10 min 15 min 5min 100 sec 50 sec

Radiation timestep 180 min 180 min 60min 15 min 5 min

Convection timestep 5 min 7.5 min 2.5min 100 s 50 sec


Model Physics Comparison GL NAE UK4

Clouds :based on Smith(1990)

Prognostic cloud scheme based on conserved variables TL and qT and a sub-grid scale probability distribution of these variables, to derive cloud amounts and water contents assuming critical relative humidity. The scheme is modified such that only water clouds are defined from TL and qT and a sub-grid probability distribution. Ice water content is determined by the mixed phase microphysics scheme with ice cloud fraction calculated diagnostically from ice water content.

Plus additional parameterisation to derive cloud area as well as volume

Radiation :based on Edwards, Slingo (1996)

Rigorous solution of the two-stream scattering equations including partial cloud cover. Full treatment of scattering and aerosols. Consistent treatment of cloud radiative properties in solar and thermal regions of spectrum. Ice crystals are treated as non-spherical. Some simplifications are made in the UK configuration for speed of computation.

Treatment of aerosols using a background aerosol climatology (Cusack et. al. 1998)

Inclusion of biogenic aerosols

Radiation parameterisation to take account of slopes in orography.


Model Physics Comparison

GL NAE UK4

Boundary Layer : Lock et al, 2000 and Martin et al, 2000

1. Stability dependent surface exchanges based on Monin-Obukhov length. Boundary layer mixing scheme diagnoses 7 boundary layer types with non-local mixing and explicit entrainment parameterisation for unstable boundary layers, local Richardson number-based scheme used for stable layers.

2. Additional shear driven turbulent BL (type 7)

3. Revisions to BL & surface transfer

-Non-local mixing for momentum (Brown et. al. 2007)

- Surface scalar transfer over sea – adjustment to roughness length to reduce exchange coefficient for heat and moisture at high wind speeds in line with observations. Effect of salinity in reducing saturation vapour pressure included (Edwards (2007)

Changes to depths of non-local mixing in decoupled Sc boundary layers (8B BL scheme).

Stability functions:

(A) Sharpest over the sea

(B) Long tails

(C) Louis func. near surface & sharp above

AB AC C


Model Physics Comparison

GL NAE UK4

Gravity Wave Drag : Webster et al, 2003

Flow blocking scheme that simplifies diagnosis of hydrostatic gravity waves and low level drag based on Froude Number.. Updated to use GLOBE orography.

Including a parameterisation of orographic roughness (Milton & Wilson, 1996)

()

Precipitation : Wilson and Ballard, 1999

Bulk microphysics scheme with one ice variable and explicit calculation of transfers between vapour, liquid and ice phases. Cloud microphysics with an additional ice variable and explicit calculation of transfers between vapour, liquid and ice phases.

Prognostic rain (3D advected)

Improved microphysics – including changes to auto-conversion, consistent sub-grid model to represent grid-box heterogeneity, two particle distribution of ice, calculation of ice fall speeds, changes to deposition and sublimation of ice, and revised ice nucleation, (3C microphysics)



Convection: see Gregory & Rowntree, 1990; Gregory & Allen, 1991;

Grant 2001; Grant & Brown,1999

1. Modified mass flux scheme including changes to diagnosis and triggering of deep and shallow convection

2. Plus convective momentum transport.

3. Plus representation radiative effects of anvils

CAPE closure based on humidity threshold. Default timescale for removal of CAPE in the column and hence mass flux triggering convection at base is set to 30 minutes in global and NAE.

CAPE-dependent CAPE closure; Timescale linear with large CAPE

Small CAPE values (shallow convection): Timescale tends to a minimum limit. Convection works as normal.

Large CAPE values (deep convection): Timescale grows linearly with CAPE. Convection inhibited and microphysics takes over.

W-based CAPE closure – CAPE depends on vertical velocity above a given threshold (decreases with increasing w)

Adaptive detrainment for deep & mid-level convection (Maidens & Derbyshire, 2007)



Land Surface: Essery et al, 2002

MOSES II: Additionally includes "tile" scheme with separate surface-atmosphere fluxes calculated for each surface type present in a given gridbox.

Soil moisture is initialised by a nudging scheme (Best & Maisey, 2002) with corrections derived from increments to screen level T and Q in a separate 'screen VAR' analysis.

(Away from UK)

Soil moisture is reset daily using data from the MOSES-PDM Model (Smith et al., 2004). This is run independently for a domain slightly larger than the UK and Ireland. It includes MOSES II physics and is driven by observed precipitation and cloud.

Revision to surface albedo ancillary files based on MODIS observations.

Anthropogenic heat source on tiles. (Urban heat island)


40y performance improvement

RMS surface pressure error over the NE Atlantic


Parallel Suites

•Near real time running to test the next version of the Operational NWP Suite.

•Essential Final Validation of a Science Change

•Provides Trouble-free!!! Mechanism for installing changes.

•Runs on the Backup System

•Used to implement significant structural or scientific changes

•Each Parallel Suite is ~4 weeks long.

•~ 3 time per year. => Packaging different scientific changes together.


2009 Scheduled Changes

PS20 Nov 2008

• Physics updates in all configurations (surface, radiation and microphysics)

• Revised global Covariance statistics

• “MOPS” cloud assimilated in NAE & UK4 via VAR rather than by nudging

• Significant revision to Visibility parameterisation in UK4

• Inclusion of Windsat in global assimilation

• Use of SEVERI Clear sky radiances in NAE & UK4

PS21 May 2009

• Production System to IBM p6 & full operational support for MOGREPS

• Introduction of Trial UK1.5km suite

PS22 Aug 2009

• 70 levels in global and NAE

• Seasonal Model Upgrade GloSea4

PS23 Nov 2009

• 25km global with matching increase in 4DVAR resolution

• 60km global EPS and 16km regional EPS


Our longer term Forecasting Strategy…

1km

2009 2011/12 2014

1.5km UK 1.5km UK Ensemble

•25km Global

•12km NAE

•16km NAE Ensemble

•16km Global

•(12km NAE)


Now

•40km Global

•12km NAE


4km UK


Climate models, coupling & seamless prediction


GloSea4

• Expected to become operational in April 2009

• Later to be: (1) integrated with our decadal system and (2) implemented in Korea and India

• Main drivers:

•Seasonal to decadal as an adaptation tool to climate change

•Bridge between NWP–Climate to facilitate model development

•…. and improve seasonal forecasts!


Monday

ADA(NWP) + ODA

Weekly ICs

N forecasts (6-month)

Archiving/post-processing

GloSea4 prototype:System design/infrastructure

Tue - Sun

ADA(NWP) + ODA

N forecasts (6-month)

Archiving/post-processingFC

ST

Every week

• Pull together last 4-weeks fcsts/hcst

• Post-processing

• Products / web displayEvery day

‘Weekly’ ICs

N forecasts (6-month): Y-years, M-members

Archiving/post-processingHC

ST

Off line

• Continuous ODA (e.g. 1989 – 1998)

• ‘Weekly’ Atmos + land.surf. Reanalysis data


GloSea4 operational implementation

April 2009 Future upgrade

System configuration

Fcst: 30 membersHcst: 15-yr (most recent years from ERA-

Interim); 10 members

Fcst: 30 membersHcst: 20-yr; 10 members [cost incr.: x 1.3]

Model configuration and resolution

HadGEM3 N96L38+ORCA1L42 HadGEM3 N96L60+ORCA1L42 [cost incr.: x 1.6 (with same

timestep)]

Initialization Atmosphere: reconfiguration of NWP / ERAOcean: DA: fix problems and improved

covariances; improved mixing schemes [TKE]

Sea-ice: Include assimilation

Coupled initialization as described in Anderson et al. (2008).

Replacement of NEMO-OI by NEMOVAR

Representation of uncertainties

ICs: lagged approachModel:RP + SKEB2

Improvement of SKEB2

Post-processing - GloSea3 calibration and display system- New calibration methods- Complete MO-EPS Integrated web display

system

- Improvement of new calibration methods

- Improvement of integrated display system

Computing cost ~1.5% ~3%


HadGEM family• HadGEM1

•IPCC AR4 (CMIP3)

• HadGEM2-AO

•ENSEMBLES

•Physical model for HadGEM2-ES

• HadGEM2-ES

•IPCC AR5

• HadGEM3-AO

•Under development

•Uses NEMO and CICE models

•Basis for GloSea4


Our Approach

• To separate physical and earth system model development

HadGEM2: Quantification of Earth System feedbacks and understanding the uncertainty associated with Earth System processes.

HadGEM3: A model that is ‘fit’ for purpose for application across 1-Day -Seasonal-Decadal-Centennial scales and for regional prediction

• To better measure improved performance

• To explore the value of enhanced resolution through collaboration

• To exploit the Unified Modelling Structure towards a seamless prediction system


HadGEM2-ES

• The climate model now includes major earth system feedbacks (carbon cycle, chemistry, aerosols)

• It is designed to run the major scenarios for IPCC 5AR

• Hadley Centre’s “standard” climate model.

• Not the last word on Earth System Modelling at the Met Office

• Not HadGEM2-ES+

• More complex chemistry/aerosols? Nitrogen cycling

• Research tool

• Not HadGEM3-ES

• New ES components (probably QUESM-based)


HadGEM2-ES Components

Fully coupled Earth System Model

• Atmosphere, ocean, sea-ice, land surface

• HadGEM2-AO + new hydrology scheme (wetland methane)

• Land ecosystems: dynamic vegetation, soil C

• TRIFFID, RothC

• Ocean ecosystems: NPZD, diatoms, non-diatoms,

• Diat-HadOCC

• Aerosols: Sulphate, BC, OC, dust, sea salt

• Current aerosol scheme, with some improvements

• Tropospheric chemistry: ozone, methane, oxidants

• UKCA


HadGEM3-AO Development

• First phase completed March 2008

•Prototype coupled model including NEMO (ocean) and CICE (sea ice) components

•Major assessment of model results based on objective assessment criteria

• Next phase: MORPH3 (Model for improved Regional Prediction – HadGEM3)

•Started September 2008

•Completion March 2010


Seamless Model Assessment

• Seamless Model Assessment is the exploitation of the seamless nature of the Unified Model across space and timescales to assess and improve the simulation of processes within the model

• Understanding systematic errors

• Shared use of detailed diagnostics

• Cross fertilization of physics development

• Traditional boundaries between Weather and Climate prediction are artificial

• Initial value component to seasonal to decadal prediction

• Many key systematic climate errors are common to short range integrations with the same physical model run from well balanced initial states


Met Office physical models 2009

Regional NWPNAE

(2 days)

Global NWP

(6 days)

THORPEX(15 days)

Seasonal

(GloSea4)

Decadal(DePreSy

s)

Centennial

Regional Climate

Horizontal Resolution

12km 25km – N512 60km – N216 135km – N96

250km – N48

135km – N96

25km

Vertical Resolution

70 L 70 L 70 L 38L 19L 38L 38L

Atmos Physics

HadGAM2 + upgrades

HadGAM2 + upgrades

HadGAM2 + upgrades

HadGAM2 + upgrades

HadAM3 HadGAM2 HadCM3

Atmos. data assimilation

4D-Var (3h cycle) MOPSLH nudging

4D-Var Reconfigured Global Analysis

Reconfigure ERA-40

Relaxation to ERA-40

N/A N/A

Soil moisture initialisation

Reconfigured daily from Global soil moisture

Soil nudging Global “nudged” analy.

UM analysis.

N/A N/A N/A

Ocean Persisted SST Persisted SST Persisted SST anomalies.

NEMO (1.25°, 1/3 ° tropics)

HadCM3 (1.25°)

HadGOM1 + upgrades (1°, 1/3° tropics)

Driven by HadCM3

SSTs

Ocean data Assimilation

N/A N/A N/A FOAM – Met Office global Ocean data assimilation

4D, multi-variate OI of salinity, sub surf.T ,SST’s.

N/A N/A

Ensembles

MOGREPS 24 member ETKF 16km/70L

MOGREPS 24 member ETKF (Bowler(06)) N216/70L

24 member ensemble from MOGREPS perturbations

40 member 16 member (initialised on different days & seasons).

HadCM3 QUMP PPE ensemble


Model Hierarchy

• Exploit the Unified Model system to develop a hierarchy of models with “common” controlling processes

• Use all relevant time and spatial scales for evaluation of key mechanisms of climate change compared to observations

Instead of a flagship model, we need a reference model. For particular purposes, we need lower/higher resolution or more/less complexity.

This is not a new idea. The challenge in implementing it is traceability.

© Crown copyright Met Office© Crown copyright Met Office

A prototype model hierarchy

Ultra high resolution• ‘Global’ Atmosphere (N320, CASCADE, …)• Reduced domain (RCM, NAE, CAM, UK4,

1km)

Seasonal/Decadal• Global OAGCM (N216,0.25° L85)• Run for multidecades on joint supercomputer• Balanced/tuned climate NOT required

Centennial/Ensembles• Global AOGCM (N144/N96,0.5°/1° ?)• Multiple multicentury runs or S2D ensembles• Needs balanced/tuned climate simulation

Earth System• Global AOGCM (N48, 2°)• Large multicentury ensembles, rapid

response• Needs balanced/tuned climate simulation

Simple model• EMIC, EBM, statistical emulator, empirical f/c

model, statistical downscaling, …• Producing pdfs, Low prob events, rapid

response

Targeted expts with specific processes

A few scenarios with large range of processes

QUMP

Complexity-QUMP

Unrestricted processes (parameterized)

Complexity & ensembles

Re

so

luti

on

Met R&DCASCADE, Earth Simulator Centre

UJCCHiGEMseasonal

Model development teams

QUEST

QU

MP

, M

itig

atio

n,

Da

ng

ero

us

CC

SM

A +

sp

eci

alis

ts

Groups involved

“Physical traceability barrier”

MOHC specialists, academics

Operational Seasonal and Decadal Forecasting


UM licensing and collaboration


UM licensing

• The Met Office Unified ModelTM is available for research, operational and commercial use

• For research use, an abstract and annual reporting is required

• For operational use, the UM suite is provided for a one-off and annual licence fee

•annual licence fee refunded upon satisfactory auditing of an R&D contribution to an agreed science plan


Intellectual Property Rights

• UM IPR remains with the Met Office

• Licensees must acknowledge the Met Office using “Met Office Unified ModelTM ” in publications etc.

• Background IP remains with the Met Office and foreground IP shall be assigned back to the Met Office


Operational & public duty use

• Operational licensees must:

•Be a recognised NMS according to WMO or have a contract to supply NWP data to the NMS

•Have no restrictions from UK international policy

•Possess the capability to run the UM effectively

•Add value to existing UM output

•Use the model for public duty use only


UM operational users

http://www.bom.gov.au/bmrc/


Collaboration projects

• NWP evaluation & diagnostics: 8• Model parameterisations & processes: 5• Obs. processing, QC & data assimilation:

5• Improved use of satellite data: 5• Ensemble forecasting: 3• Post-processing & impact forecasting: 1• Wave modelling: 1• Seasonal forecasting: 1• Global ESM evaluation & diagnostics: 3• RCM evaluation & diagnostics: 3• Infrastructure (FLUME, coupling, UKCA): 3


The ACCESS modelling system


OASIS

UM

CICECSIRO OC

UKCACABLE

AusCOM (MOM4)

LPJ

VAR

OPS

BODAS

OBS

Assimilation (?)

NWP

ACCESS modules


MetDB

OPS

VAR

UM

BoM Obs.

ODB (ECMWF)

OPS

VAR

UM

MARS

Met Office ACCESS

MetDB

UKMO Obs Observations

ObservationData-Base

Met. Archive Retrieval

Unified Model

VariationalAssimilation

Observation


ACCESS

• Implementation phase 2007-2008

• SH MSLP verification 22 May – 30 Sep 2008:


Development of east coast low - regional models

Starting analysis

Verifying analysis

ARFS

LAPS


ACCESS climate prediction

• Coupling of CABLE, CICE and AusCOM to UM

• Aim to participate in AR5

• Coupled model complete Mar 2009

• Tuning complete end 2009

• Planned climate resolution:

• atmosphere N96, 38L,

• ocean approx 1 degree, 46 L

• Atmosphere to be HadGEM2-A with PC2

• Tests with SCM and AMIP type experiments; http://www.accessimulator.org.au/workshop/oct08.html

http://www.accessimulator.org.au/workshop/oct08.html


Questions and answers


RCM INPE collaboration

• The Hadley Centre climate model famously predicts warming, drying Amazon climate and “Amazon die-back”

• Other climate models project drying, but not as much

• Climate change may increase risk impact of deforestation through fire leakage

• All work so far done in global model. Now collaborating with INPE / CPTEC on improved regional climate modelling (funded by UK Foreign and Commonwealth Office) – Richard Betts, Jose Marengo, Gillian Kay, Chou Sin Chan

• Uncertainties in regional climate model simulations

• Incorporate vegetation feedbacks in regional climate model


Impacts on ecosystems and role of climate feedbacks

Difference in forest area 2080-2000

Precipitation difference 2080 - 2000 (mm day-1)

Precipitation difference due to forest feedbacks (mm day-1)


Uncertainties in climate projections: changes in S American precipitation in Hadley Centre ensemble


Projected increase in fire risk due to climate change

2020s 2080s

Proportion of climate model simulations projecting “high” fire risk (McArthur fire danger index)

Ensemble of simulations with HadCM3 climate model

Golding and Betts (2008) Glob. Biogeochem. Cycles


Satellite collaborations Met Office - CPTEC

• Atmospheric motion information from geostationary imagers e.g. SEVIRI

• Renato Galante planned visits to Met Office March 2009 and November 2009. The focus is on tropical convection.

• GPSRO

• Luiz Sapucci visit to Met Office April-May 2009

• Tropical humidity, land surface temperature

• Stephen English is planning to visit CPTEC March-April 2009 to compare humidity analyses, especially in the tropics, to initiate collaboration in assimilation of moisture observations in the tropics.

• Also a comparison of land surface temperature and methods for assimilating ATOVS is planned.


Multi-variate Statistical/Dynamical forecast probabilities for rainfall terciles

• Example forecast NE Brazil rainfall for the wet season issued since 1987 by the Met Office

• Produced with GloSea3 + statistical model using indices of Tropical Atlantic and Pacific SST as predictors

• Contribute to Regional Climate Outlook Forum for NE Brazil organised by FUNCEME and others

• Forecasts are also available on our websitehttp://www.metoffice.gov.uk/science/specialist/seasonal/

http://www.metoffice.gov.uk/science/specialist/seasonal/

Date post:	13-Dec-2015
Category:	Documents
Upload:	laura-mitchell
View:	215 times
Download:	0 times

© Crown copyright Met Office Weather and climate modelling workshop, INPE, Cachoeira Paulista Met...

Documents