COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – WG1 Priority Projects and New Activities Deutscher Wetterdienst, D Offenbach, Germany

COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – Project: Km-Scale Ensemble-based Data Assimilation (KENDA) Aim:Development a novel ensemble-based data assimilation system for the convective scale (or km-scale, i.e. 1 – 3 km model resolution)‏ and to show that it works scientifically and gives a systematic positive impact, in particular in convective situations, but also for low stratus conditions and near steep orography (local wind). Motivation:Need to provide initial conditions for the convective-scale ensemble forecasting using advanced data assimilation methods

COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – Project: Km-scale Ensemble-based Data Assimilation (KENDA)‏ Task 1:General issues in the convective scale and evaluation of COSMO-DE-EPS Purpose: Guides the subsequent decision how the resources should be splitbetween EnKF and SIR (COSMO-NWS and universities); part of the learning process Task 1.1: General issues of interest are the following: ·main disadvantage of LETKF: assumes Gaussian error distributions  investigate non-Gaussianity by means of O – B statistics (convective / larger scales, different forecast lead times: 6, 3, 1 hr): provides an upper limit estimate of the non-Gaussianity to deal with ·LETKF produces model states with linear combinations of model perturbations  examine influence of non-Gaussianity on the balance of such model states: also provides an upper-limit estimate on unphysical behaviour of linear combinations in the case non-Gaussian PDF ·assess importance of km-scale details versus larger-scale conditions in the IC Remote. DWD would supply the data to be evaluated. 0.5 FTE. Task 1.2.: Evaluate COSMO-DE-EPS for applicability for km-scale EnDA. Features of interest: ·ensemble size ·spread in the forecast range up to 3 hours ·drift of solutions ·indication of non-Gaussian distributions Already allocated at Uni Bonn. DWD supplied the data. 0.5 FTE. Milestone: Dec 2007: Initial evaluation conducted, feedback to COSMO-DE-EPS Resources required totally: ca. 1.0 FTE FTE require solid background (Masters deg.) in meteorology, physics and / or mathematics, good knowledge of statistics and comprehension of dynamic meteorology Milestones: ·Dec 07: Initial evaluation conducted, feedback to COSMO-DE-EPS (0.3 FTE, Uni Bonn)‏ ·…..

COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – Project: Km-scale Ensemble-based Data Assimilation (KENDA)‏ Task 2:Set-up of an ensemble data assimilation test bed (2 options to implement analysis step for EnKF:) separately outside (or integrated inside model)‏ Task 2.1: COSMO model code (0.4 FTE): –writing out obs increments (including flags and obs. errors) to NetCDF files (for EnKF code (option 1 only) and / or diagnostics)‏ –modifications for ensemble runs (related to GRIB, etc.; GRIB2 ?)‏ ( use observation increment calculations of the nudging scheme for existing obs types; for novel observations (e.g. 3D radar reflectivity), additional work (FTE) will be required) Task 2.2: Analysis step code (EnKF / 3DVAR) (0.3 FTE): –adaptation of global analysis code (ETKF for GME) to COSMO model grid –reading of observation increments (for option 1 only)‏ Task 2.3: COSMO model code, required for implementation option 2 only (0.5 FTE): –implementation of EnKF in the COSMO model –parallel forecasts (organisational overhead, communication)‏ Task 2.4: Extension of experimentation system (0.1 FTE) Task 2.5: Ensemble-related diagnostic tools (0.2 FTE)‏ Start with diagnostic tools of COSMO-DE-EPS, complement input (NetCDF obs incr with flags instead of obs and model forecasts separately), introduce selection criteria dep. on flags Resources required totally: 1.0 – 1.5 FTE. FTE require good programming skills & solid maths background (Masters deg in related field) Proposal: 4-eye principle, having 2 people (1 DWD (?, depends on the approval of DWD strategy), 1 other person (ca. 0.7 FTE) working together on whole task 2

COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – Project: Km-scale Ensemble-based Data Assimilation (KENDA)‏ Task 3: Evaluate and optimise LETKF (needs to be detailed further)‏ Issues: Model perturbations, covariance inflation, localisation (multi-scale DA ?), convection initiation (warm bubbles, LHN), etc. Resources required totally: 3 Task 4 + (?): Develop and evaluate the SIR filter 1.define norm (estimate most likely members by measuring ‘distance’ obs.)‏ 2.develop resampling strategy to construct a new ensemble reflecting posterior pdf 3.evaluate performance of SIR scheme Required Resources: - required this year:X FTE - able to contribute: X FTE - required in total: X FTE (up to 201X)‏ - Project Leader:Christoph Schraff (ad interim)

COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – Project: Retrievals for Nudging: 1D-Var for Satellite Radiances Aim:Operational assimilation of ATOVS, SEVIRI, & IASI data for LM Motivation:Use of satellite radiances especially over data-sparse areas over sea has the potential to provide essential information to improve forecasts. Tasks:(remaining)‏ T3, T4: Implementation for ATOVS : take up again at IMGW (0.25 FTE)‏ AIRS / IASI: continue (Eumetsat fellow, DWD, 0.4 FTE)‏ T5: cloud and rain detection, bias correction (review + refinement)‏ Are interrelated, iterative approach, needs to be re-tuned for each model application ATOVS (IMGW: 0.2 FTE), IASI (DWD, 2 FTE, FTE for ATOVS)‏ T6: background errors perform studies based on different background errors based on NMC or EnKF covar. (DWD: 0.05 FTE, ARPA-SIM: 0.1 FTE)‏ use only SEVIRI WV-channels over land, WV and IR channels over sea use ATOVS over sea only T7: stratospheric first guess: Option 1: simplest possiblity: use ECMWF forecasts, works already for GME: implement this first because scientifically solved. Use as benchmark for option 2 Option 2: extrapolation with statisical regression

COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – T7: stratospheric first guess: Option 1: simplest possibility: use ECMWF forecasts, works already for GME: implement this first because scientifically solved. Use as benchmark for option 2 Option 2: extrapolation with statistical regression T8: validation of retrievals with radiosonde: to be continued (DWD: 0.05 FTE, ARPA-SIM: 0.1 FTE)‏ T9: tuning, case studies (in iteration with T8) (DWD: 0.4, ARPA-SIM: 0.1, IMGW: 0.1 FTE)‏ tune observation errors, horizontal thinning, background errors for iterated retrievals, nudging weights validation and case studies (ad hoc)‏ evaluate forecast scores on test periods Aim:Show that assimilation of ATOVS and SEVIRI COSMO model works scientifically and that it gives a positive impact (also systematically) Use nonlinear observation operators within nudging framework IASI: technical implementation and first case studies Required Resources: - able to contribute:2.2 FTE ( 1.2 DWD, 0.3 ARPA-SIM, X IMGW) - Project Leader:Reinhold Hess (DWD) Open issue: Contribution of HMC for specification of observation error covariances Needs to be detailed (method, data basis)‏

COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – DWD: (monitor of COSMO-DE with respect to LHN, trouble shooting if necessary)‏ –further diagnose convection in the model without and with LHN, e.g. vertical structure (wind field) and environment of precipitation cells (moisture convergence), better understand LHN and how the model produces convection by itself –extend forecast range with benefits from LHN in cases of organized convection –improve vertical distribution of LHN increments, possibly use of 3D reflectivity –(horizontal filtering)‏ –improve quality control (e.g. bright band), extend radar domain (F, CH, NL, B)‏ MCH:– Continue test suite into fall and winter, if positive include LHN in pre-op suite, –operational introduction of LHN next Jan., validate test suites on 2.2 km; –extension of quality flags / measures, extend radar domain 1.1Use of Radar Data 1.1.1Latent Heat Nudging Klaus Stephan (DWD), Daniel Leuenberger (MCH) 1.1.2Use of radar-derived rain rates by a 1DVAR approach Virginia Poli (ARPA-SIM) ARPA-SIM: PhD for 2 years. Derive temperature and humidity profiles from radar rain rates by a 1DVAR approach (compared to LHN: different size and vertical distribution of T, q increments)‏

COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – Status become unclear data pre-processing: implement interpolation from polar coordinates to Cartesian grid using a Cressman weighting scheme (as in ADAS / ARPS) scheme (or a more simple approach) already done Set up nudging of retrievals and compare SAR + nudging with VRDAS variational radar data assimilation system in the ARPS model. Choose the better and less computing demanding option. Interest to work further on the quality control of other available radar data in COSMO D Simple Adjoint Wind Retrieval Jerzy Achimowicz (IMGW) DWD: –correct height assignment error in German VAD pre-processing, if done, re-do monitoring and assimilation for test periods –monitoring of radial winds –study literature to decide on method how to use radial winds: e.g. 3D SAR (WP 1.1.3)‏ MCH: test suite for use of operational assimilation (at least) Swiss VAD + SODAR 1.1.4Doppler winds and wind profiles Michael Buchhold (DWD); Christophe Hug (MCH)

COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – GPS tomography (MCH (50 % FTE): –(vertical profile of) bias correction of GPS tomo retrieval (against radiosonde (or model ?)‏ –assimilation of humidity profiles into COSMO (first step / proof of concept)‏ –derive q v from N wet, (use T and p from model, assuming no errors)‏ –direct nudging of q v profiles –assure that the vertical variability of the model is not destroyed by applying vertically smoothed observation increments –check for using 1DVAR approach to derive temperature and humidity profiles instead –impact on analysis and forecast quality, tuning of structure functions etc. –further work on tomography e.g. by ETHZ (new PhD (Perler)) and by GFZ Potsdam / Uni Leipzig ZTD-derived integrated water vapour: implementing anisotropic structure functions ? Implement in COSMO-DE, make more stations available, use for quality control of other data, impact studies 1.3Production and Use of Cloud Analysis Christoph Schraff (DWD)‏ 1.2Multi-Sensor Humidity Analysis (incl. GPS-obs) Daniel Leuenberger (MCH) produce profiles from radiosonde, synop, ceilometers, etc. use MSG-NWCSAF cloud type product as cluster analysis skipped due to insufficient resources --> anybody interested to do it ?

COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – Assimilation of Screen-level Observations PBL Analysis Jean-Marie Bettems, Christophe Hug, Andre Walser (MCH), Andrea Rossa + collaborators (ARPA-Veneto), NN (ARPA-Piemonte ?) Work plan: despite first case studies, all tasks still need to be done 1. a: careful analysis of Schraff low-stratus cases, find critical issues on assimilation of 2-m humidity (Walser);  use this to refine priorities within this W.P. b:estimation of PBL height and extrapolation of 2-m obs to 10 m using standard obs (Rossa, Rotach). a: improve selection of representative stations (Rossa) (statistical approach (conditional verif), or topological approach). b:improve observation screening of representative stations (Rossa) (QC, VERA, stablity criteria). improve meteorological pre-processor to extrapolate obs (Milelli ?, Buzzi). a:improve vertical structure functions, e.g. by use of PBL height (Milelli ?). bimprove lateral structure functions (Hug, Bettems) (mother-daughter alg. (ADAS), statistical approach, gridded data). c:address type of balance satisfied by the wind in complex topography (Hug, Bettems); also consider relation with pressure field Resources unclear: MCH: Hug, Walser 6 mo, Bettems few mo, ARPA-V: Rossa + others 85 % FTE (only validation); ARPA-P: unclear

COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – use of new generation of satellite instruments (METOP: AMSU-A, ASCAT winds, GRAS T,q profiles; IASI L2 products (T,q retrievals) when available)‏ continue to investigate and refine LETKF investigate incremental digital filter initialisation to be applied to interpolated 3DVar analysis for 2.8 km COSMO-IT (as an alternative to nudging)‏ 1.5Assimilation of Scatterometer Wind Heinz-Werner Bitzer (MetBw), Alexander Cress, Christoph Schraff (DWD) 1.73DVAR / LETKF for the mesoscale Massimo Bonavita, Lucio Torrisi, Antonio Vocino (CNMCA) testing & tuning in case studies and longer test periods, e.g. with respect to vertical (and horizontal) structure functions, thinning of obs, degree of explicit geostrophic balancing, …. replace / complement AOF by NetCDF interface (input & output)‏ 1.6Evaluation / Monitoring / Tuning of Nudging Christoph Schraff (DWD)

COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – Task: use of a parameterised regression between 2-m temperature and soil moisture as a gradient in the minimisation of the cost function in the variational scheme  No extra model forecast integrations required any more, unlike in current SMI scheme validate method for all near surface parameter against observations, if results good, introduce operationally in COSMO- EU / -DE (implementation in GME is being prepared, testing should be finished in Q4 2007)‏ Possible improvements, no firm plans: –use also RH 2m observations in the cost function (after improving T2m, RH2m diagnostics within the COSMO model)‏ –analyse all 5 soil layers separately (currently, L1–3 & L4–5 are aggregated to top and bottom layer) –inclusion of precipitation analysis when good product is available –improve model error statistics (‘Q-term’ to update B) (ARPA-S ?)‏ –update physiographic surface parameters (using MODIS data) (ARPA-SMR)‏ 1.8.1Soil Moisture Initialisation Martin Lange (DWD) 1.8(Sub-) Surface Analyses

COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – Use of Lake Temperature Analysis Dimitri Mironov, Michael Buchhold, Christoph Schraff (DWD)‏ flag SST where no observation used for SST of lakes derive surface temperature from bulk lake temperature, nudge prognostic bulk temperature by using difference ‘SST-observed’ minus model-derived surface temperature evaluate long test periods Merge latest DWD snow analysis modifications with new software get ccess to additional in-situ observations (currently DWD data can not be decoded)‏ Interpolation with altitudinal gradient –more realistic than Cressman interpolation over steep topography, but enough observations with snow must be present –use gradient-interpolation to identify bad observations –merge gradient-interpolation with cressman analysis, e.g. with weighted mean Use partial snow cover in COSMO/TERRA Use EUMETSAT SAF snow albedo in COSMO (Introduce a more sophisticated snow model) Snow Cover Analysis Jean-Marie Bettems (MCH)

COSMO General Meeting, Athens, 18 – 21 Sept new list of activities WG1 List of Activities 2007 – netCDF obs 3DVar netCDF feedback Cosmo model BUF R bufr 2 wmo_buf r WM O BUF R bufr2netcdf verification netCDF 2 ODB monitoring IT section NWP section SKY / archive ~ 1 by 1 converter simple + portable applicable to WMO or non-WMO BUFR standard WMO templates, i.e. unique descriptors + dimensions of elements + code tables unique BUFR format for each obs type can keep AOF as alternative data input as long as needed Advantages of NetCDF: widely used and portable a variety of software exists to plot, analyse and evaluate the data. DWD plans: envisaged set-up observation formats, pre- and post-processing any kind of DWD will switch to NetCDF by March 2008 thereafter, DWD will no longer support AOF interface Under discussion at DWD