Presentation on theme: "The Operational Data Assimilation System"— Presentation transcript:
1 The Operational Data Assimilation System Lars Isaksen, Data Assimilation, ECMWFContent of talkOverview of the operational data assimilation cycleComputational issuesObservations used by the ECMWF Assimilation SystemMulti-incremental 4D-VarWhy is 4D-Var performing better than 3D-Var?Recent improvements of ECMWF’s Assimilation SystemNear future data assimilation implementations
2 Data assimilation system The observations are used to correct errors in the short forecast from the previous analysis time.Every 12 hours we assimilate 7 – 9,000,000 observations to correct the 80,000,000 variables that define the model’s virtual atmosphere.This is done by a careful 4-dimensional interpolation in space and time of the available observations; this operation takes as much computer power as the 10-day forecast.
3 The operational configuration at ECMWF Deterministic model: T1279L91 (~16km)Outer loop of 4D-Var T1279L91 and inner loops T159/T255/T255 (~125km/80km/80km)EPS target resolution T639L62 (to 10 days) and T319L62 thereafterWave model (25km and 36 directions)Implemented in operations 26 January 2010
4 Observation pre-processing for 0000UTC main cycle Extract data for 12h period UTC70sec (min. 8x1PEs)Pre-process satellite data. Cloud clearing. Scatterometer winds.340sec (min. 16x1PEs)
5 Analysis: trajectory, minimization and update 4320sec (3072PEs) Analysis and forecast for 0000UTC main cycleBUFR to ODB sec 4x(8-16PEs)Fetch background forecast 275sec 2x(1PE)430s1110s260s880sAnalysis: trajectory, minimization and update 4320sec (3072PEs)270s820s490sSurface analysis sec 4x(1PEs)10 day forecast t-steps sec (3072PEs)(or 15h fc for cycling: 260sec)
7 Operational schedule Early delivery suite introduced June 2004 02:0003:3014:0015:3012h 4D-Var, obs 21-09Z12h 4D-Var, obs 09-21Z04:0016:0006 UTC analysis18 UTC analysis3hFC6h 4D-Var21-03Z3hFC6h 4D-Var9-15Z03:4004:4015:4016:4012 UTC analysis (DA)00 UTC analysis (DA)T day forecast18:05T day forecast06:0551*T639/T399 EPS forecastsDisseminate51*T639/T399 EPS forec.DisseminateDisseminateDisseminate05:00Disseminate06:3517:00
8 Model/4D-Var dependent QC Data extractionBlacklistingData skipped due to systematic bad performance or due to different considerations (e.g. data being assessed in passive mode)Departures and flags available for further assessmentCheck out duplicate reportsShip tracks checkHydrostatic checkThinningSome data is not used to avoid over-sampling and correlated errorsDepartures and flags are still calculated for further assessmentModel/4D-Var dependent QCFirst guess based rejectionsVarQC rejectionsUsed data IncrementsAnalysis
9 Conventional observations used SYNOP/METAR/SHIP:MSL Pressure, 10m-wind, 2m-Rel.Hum.DRIBU: MSL Pressure, Wind-10mRadiosonde balloons (TEMP):PILOT/Profilers: WindWind, Temperature, Spec. HumidityNote: We only use a limited number of the observed variables; especially over land.Aircraft: Wind, Temperature
10 Satellite data sources used in the operational ECMWF analysis 13 Sounders: NOAA AMSU-A/B, HIRS, AIRS, IASI, MHS5 imagers: 3xSSM/I, AMSR-E, TMI3 Scatterometer sea winds: ERS, ASCAT, QuikSCATGeostationary, 4 IR and 5 winds4 ozone2 Polar, winds: MODIS6 GPS radio occultation
11 Significant increase in number of observations assimilated Conventional and satellite data assimilated at ECMWF
12 The observation operator Observations are not made at model grid pointsSatellites often measure radiances, NOT temperature and humidityWe calculate a model radiance estimate of the observation to enable comparison.This is done with the ‘observation operator’ H.H may be a simple interpolation from model grid to observation locationH may possibly perform additional complex transformations of model variables to ‘radiance space’ for satellite data.ModelT,u,v,qHModel radiancecompareObservationSatellite Radiance
13 The variational method allows model radiances to be compared directly to observed radiances Enables use of advanced observation operatorsModelT and qObservationSatellite RadianceHModelRadiancecompare
14 Jb: Ensures that the background model fields are adjusted meteorologically consistently in the region close to the observation locationIncrements due to a singleobservation of geopotential height at 1000hPa at 60N with value 10m below the background.
15 Increments for a single observation of geopotential height at 1000hPa. Jb: The Balance Operator ensures the height and wind field balance is retainedwind increments at 300hPawind increments at 150 metre above surf.Increments for a single observation of geopotential height at 1000hPa.Left panel: wind increment near 300hPa.Right panel: wind increment about 150m above the surface.
16 4D-Var implementation used at ECMWF All observations within a 12-hour period (~9,000,000) are used simultaneously in one global (iterative) estimation problemObservation minus model differences are computed at the observation time using the full forecast model at T1279 (16km) resolution4D-Var finds the 12-hour forecast evolution that optimally fits the available observations. A linearized forecast model is used in the minimization process based on the adjoint methodIt does so by adjusting surface pressure, the upper-air fields of temperature, wind, specific humidity and ozoneThe analysis vector consists of 80,000,000 elements at T255 resolution (80km)The size of the control vector is approx 8M.09Z Z Z Z Z
17 4D-Var incremental formulation Courtier, Thépaut and Hollingsworth (1994) In the incremental formulation the cost function J is expressed in terms of increments x with respect to the background state, x=x-xb, at initial time.and are the TL of and , linearized aroundThe innovations di are calculated using the non-linear operators, Hi and Mi .The i-summation is over N=25 ½-hour long time slots of the hour assimilation period.This ensures the highest possible accuracy for the calculation of the innovations di , which are the primary input to the assimilation!
18 Revision of operational 4D-Var algorithm Implementation:Inner/outer iteration algorithmHessian pre-conditioningConjug. Gradient minimisationImproved TL approximationsMulti-incremental T159/T255/T255These developments facilitate:Use of higher density dataHigher resolution inner loop (T255) and outer loop (T1279)Enhanced use of (relatively costly) TL physicsCloud and rain assimilationMore work has been done to improve the representation of the smallest scales in the inner loop – i.e. the analysis increments.
19 Multi-incremental quadratic 4D-Var at ECMWF T1279L91T159L91T255L91
20 4D-Var with three outer loop: efficient, accurate and allows non-linearity Analysis increments at 900hPa for temperature for each of the three minimizations.Plot shows 2009 setup. From 2010: T95T159 and T799T1279.Decreasing amplitude of increments: T95>T159>T255.Last analysis step adds small corrections where data density is highest.Note model and observation operators are re-linearized twice.T95Add T95 increment to T799 background and re-linearize M, HT159Add T159 increment and re-linearize M, HT255Add T255 increment = final T799 analysis
21 Physical processes in the ECMWF model An accurate model representation of the atmosphere is an important part of the assimilation systemThe atmosphere does not evolve in isolation, interactions between the atmosphere and the underlying land and ocean are also important in determining the weather. Ocean ice processes, ocean surface waves, land surface, soil, hydrological and snow processes are all represented at ECMWF in the most advanced operational Earth-system model available anywhere.These physical processes have smaller scales than the model grid (16 km) and are therefore represented by so-called “Parametrization Schemes” which represent the effect of the small-scale processes on the large-scale flow.
22 Outer loop model resolution is now T1279L91 Important for accurate comparison against observations T799 (25km)843,490 grid-points per levelL91L60T1279 (16km)Since January 20102,140,704 grid-points per level
23 Resolution and QC benefits analysis of extreme events Improved hurricane analyses due to Huber norm and resolution(35r3) Use of improved QC (Huber norm)961hPaOPS (35r2)980hPaHurricane Bill,20 Aug. 2009Observed MSL pressure~944hPa(36r1) High-res system T1279+T159/T255/T255945hPa
24 Scalability of IFS model – T1279 L91 Can we use our parallel computers efficiently?Scalability of IFS model – T1279 L91NodesCores
25 4D-Var is more difficult to parallelize efficiently idealT1279 modelT1279 modelT255anT1279obsT159anT1279obs=576 cores=1152 cores=1728 cores=2304 cores
26 Parallel performance is important Parallel performance is important. Scalability is not perfect for the 4D-Var analysis4D-Var T799 outer,T95/T255 inner, 91 levels on IBM system51276810241536
27 Hurricane Lili. Surface scatterometer winds Hurricane Lili. Surface scatterometer winds. An example how 4D-Var propagates information verticallyFirst guess MSL pressureAnalysis MSL pressureMSL pressure Analysis incrementsNSCAT analysisNo SCAT analysisS.M. Leidner, L. Isaksen and R.S. Hoffman ‘Impact of NSCAT Winds on Tropical Cyclones in theECMWF 4DVAR assimilation system’ Mon. Wea. Rev. 131,1,3-26 (2003)
28 ECMWF 4DVAR assimilation system’ Mon. Wea. Rev. 131,1,3-26 (2003) Hurricane Lili. Surface scatterometer winds. An example how 4D-Var propagates information verticallyFirst guessMSL pressureFirst guessMSL pressureAnalysisMSL pressureMSL pressureAnalysisincrementsNSCAT analysisNo SCAT analysisS.M. Leidner, L. Isaksen and R.S. Hoffman ‘Impact of NSCAT Winds on Tropical Cyclones in theECMWF 4DVAR assimilation system’ Mon. Wea. Rev. 131,1,3-26 (2003)
29 Surface scatterometer wind information is propagated vertically and improve the analysis. Due to flow-dependent structure functions in 4D-Var
30 4D-Var is using more a-synoptic data than 3D-Var 4D-Var is using more data from frequently reporting stations.The plots show the use of SYNOP surface pressure observations.Column height gives the number of observations available, while the black part displays those actually used in the assimilation.4D-Var SYNOP Screening3D-Var SYNOP Screening3D-Var is like 4D-Var without the time dimension. The analysis is performed at synoptic times only (0000, 0600, 1200 and 1800 UTC). Mostly only data valid a synoptic time is used.
31 4D-Var versus 3D-Var and Optimum Interpolation 4D-Var is comparing observations with background model fields at the correct time4D-Var can use observations from frequently reporting stationsThe dynamics and physics of the forecast model in an integral part of 4D-Var, so observations are used in a meteorologically more consistent way4D-Var combines observations at different times during the 4D-Var window in a way that reduces analysis error4D-Var propagates information horizontally and vertically in a meteorologically more consistent way4D-Var more complex: needs linearized perturbation forecast model and its adjoint to solve the cost function minimization problem efficiently
32 Recent revisions to the assimilation system Use of many new satellites and new instruments (will be presented by Peter Bauer tomorrow)Variational bias correction of satellite radiancesAdaptive bias correction for radiosondes and SYNOP pressure dataIncreased resolution from T799/T95/T159/T255 to T1279/T159/T255/T255More advanced Tangent Linear physics scheme in 4D-VarImproved handling of supersaturation in the humidity analysisHuber norm Variational Quality ControlWeak-constraint 4D-Var accounting for model errorAdvanced diagnostic tools to understand impact of observations on analysis/forecast: Forecast Sensitivity to Observations
33 Improved assimilation of satellite moisture data Assimilation of rain-affected microwaveFirst version (SSM/I radiances) 2005; extended to SSMIS, TMI, AMSR-E in 2007Direct 4D-Var radiance assimilation from March 2009Main difficulties: inaccurate moist physics parameterizations (location/intensity), formulation of observation errors, bias correction, linearity.4D-Var first guess SSM/I Tb 19v-19h [K]SSM/I observational Tb 19v-19h [K]
34 Humidity analysis improvements New humidity background error model:takes into account how humidity errors are affected by temperature errors in cloudy conditions (statistical relationship)extension of the humidity errors to cater for supersaturated conditions (with respect to ice)OldNewChanges in cloud cover spin-up. Monthly average zonal mean (Sep 2008) analysis minus a 15h forecast valid at the same time.
35 Improved quality control of observations Huber-norm observation cost function L1 metric far from centre, L2 metric close to centre.Robust method: a few erroneous observations does not ruin analysisAdds some weight on observations with large departuresA set of observations with consistent large departures will influence the analysisMost obs-bg innovations follow Huber-norm distribut.GaussianHuber norm (New operational)Gaussian + flat (operational)
36 How to account for model error in 4D-Var First step is a weak-constraint 4D-Var that handles biases.Q only applied in stratosphere.Cost function including model error and bias:Model error ηModel bias (cycled) ηb
37 Weak-constraint 4D-Var formulation captures part of model error biases in the stratosphere AMSU-A first guess departures reduced
38 Enhance diagnostics of the assimilation system Forecast Sensitivity to ObservationsJ is a measure of the forecast error (as defined through e.g. dry energy norm)Example:Impact of each observing system in ECMWF’s assimilation system
39 Analysis increments (mm) Soil moisture assimilation using Ext. Kalman Filter Will be implemented in June 2010Analysis increments (mm)
40 Ensembles of data assimilations Is being implemented now, first for EPS later for 4D-Var Run an ensemble of analyses with randomly perturbed observations and SST fields, and form differences between pairs of analyses (and short-range forecast) fields.These differences will have the statistical characteristics of analysis (and short-range forecast) error.To be used in specification of background errors = “errors of the day”. To indicate where good data should be trusted in the analysis (yellow shading).Also for initialization of the EPS!
41 Estimating Background Error Statistics from Ensembles of Data Assimilations (EDA) Run an ensemble of analyses with random observation and SST perturbations, plus stochastic model error representation.Form differences between pairs of background fields.These differences will have the statistical characteristics of background error (but twice the variance).AnalysisForecastxb+εbAnalysisForecastxb+ηbBackground differences
42 Why implementing Ensemble Data Assimilation? To estimate analysis uncertaintyTo improve the initial perturbations in the Ensemble PredictionTo calculate static and seasonal background error statisticsTo estimate flow-dependent background error in 4D-Var - “errors-of-the-day”To improve QC decisions and improve the use of observations in 4D-Var
43 Ongoing developments in data assimilation at ECMWF EKF for soil moisture analysisEDA to provide flow-dependent background error information to 4D-VarVertical resolution increase planned for later in 2010 (~140+ levels TBD)Long window 4D-Var: extend to 24 hour window, improve model error termImproved assimilation of cloud/aerosols/rain observationsFlow-dependent data selectionAccount for observation error correlationsModularisation of the IFS and scalability of the assimilation system
44 Summary of today’s lecture Overview of the operational data assimilation cycleComputational issuesObservations used by the ECMWF Assimilation SystemMulti-incremental 4D-VarWhy is 4D-Var performing better than 3D-Var?Recent improvements of ECMWF’s Assimilation SystemNear future data assimilation implementations