Report of the working group on Predictability, Dynamics and Ensemble Forecasting Co-chairs: Craig Bishop and John Methven Objectives The overarching objectives of the PDEF working group are: To provide a centre of scientific expertise on dynamics, predictability and ensemble forecasting that was previously not covered in the core WWRP programme; To involve both operational and academic communities in scientific research and development challenges; To support the THORPEX legacy projects, S2S, PPP and HI Weather, and the WWRP FDPs and RDPs We are currently focused on advancing science in the following 5 areas: 1. Stochastic representation of the effect of unresolved processes in numerical models 2. Construction of ensemble initial conditions 3. Interactions of diabatic processes with meso/synoptic scale dynamics 4. Assessment of multi-model ensembles and calibration techniques 5. Coupled modelling & assimilation Highlights of 2016 Perhaps the two most important contributions of PDEF in 2016 were its contributions to the PDEF/ECMWF model error uncertainty workshop and the NAWDEX field campaign. The model error uncertainty workshop has resulted in summary and guidance reports which now reside on the ECMWF web site and the NAWDEX field campaign is currently underway. Key meetings associated with these and other PDEF activities were as follows: 4-6 April, 2016, Oberpfaffenhofen, Germany: NAWDEX preparation workshop. (John Methven) 6-8 April 2016, Reading, UK: PDEF/SPARC workshop on blocking. (John Methven, Olivia Martius) 11-15 April 2016, Reading, UK : PDEF/ECMWF Model error uncertainty workshop. (Craig Bishop, Judith Berner, Mark Rodwell) 25-26 April 2016, Exeter, UK: PDEF Working Group meeting (entire working group). Also convened with DAOS working group for a session on coupled data assimilation and ensemble forecasting. PDEF WG members participated in the kick-off meeting for the High Impact Weather Project (also in the Met Office, Exeter). 6-8 July 2016, Manchester, UK. Workshop on the role of diabatic processes in weather systems at the Royal Met Soc conference on High impact weather and climate. (John Methven)
28 Aug 2016, 6 days. Grindelwald, Switzerland, Swiss Climate Summer School. (Olivia Martius) 7-9 Sep 2016, Reading, UK. YOPP planning meeting. (John Methven) 19 Sep-16 Oct 2016, NAWDEX field campaign. (John Methven) Links The PDEF working group is developing good links with the three THORPEX legacy projects and other international activities, which are summarised below: PPP As noted above, PDEF is engaged with preparations for YOPP (Year of Polar Prediction). Other concerns include: challenges of prediction at high latitudes; coupling issues for polar regions (especially stable BL); use of TIGGE & TIGGE-LAM for case studies. Plan to include additional surface flux diagnostics in TIGGE for YOPP. Using the NAWDEX field campaign month to explore the value of additional radiosondes at high northern latitudes as a consideration for YOPP. S2S Important issues of common interest include: coupled modelling & assimilation challenge; ensemble initial conditions; stochastic physics; diabatic effects. PDEF has been actively seeking researchers willing to research advanced diagnostics for the S2S data set such as diagnostics to link TC transitions with mid-latitude Rossby wave diagnostics. HI Weather The second meeting of the PDEF WG was held in conjunction with the HIWeather workshop in April 2016. Common areas of interest include: application of multi-model ensembles & calibration to improve forecast products; use of TIGGE & TIGGE-LAM for case studies. PDEF WG member Susanne Theis wrote summary guidelines on interpretation of probability forecasts from high resolution ensembles to serve as advice for the HIWeather project. WGNE joint interest in systematic errors, stochastic physics NAWDEX analysis of results to elucidate diabatic effects on model error and meso/synoptic scale dynamics. WCRP Grand Challenge on clouds, circulation & climate sensitivity storm tracks & jet stream dynamics; mid-latitude blocking & persistent weather; diabatic effects DAOS ensemble initial conditions; coupled data assimilation; value of additional observations and design of data denial experiments. Mesoscale/ Nowcasting improved treatment of model uncertainties for convective-scale prediction. CBS PDEF brings academic expertise to bear on practical problems relevant to operational forecasting, including: application of multi-model ensembles & calibration to improve forecast products; improvement of operational ensembles by improved initial conditions and better representation of model uncertainties.
Plans for 2017 Future plans were discussed at the PDEF working group meeting at Exeter and synthesized into action items. Highlights include a follow on to the ECMWF model uncertainty workshop via the 2017 WGNE workshop on systematic errors and an effort to translate some of the findings from the 2016 workshop into a recommendation for new types of coarse graining experiments. John Methven will encourage use of the NAWDEX campaign period for model error studies. Related to this, is Mark Rodwell s commitment to output tendencies (partitioned by process) from the operational ECMWF ensemble during the NAWDEX campaign. Additionally, PDEF will encourage theoretical studies in which one would attempt to predict the increase in forecast error variance associated with denying special NAWDEX observations. Munehiko Yamaguchi will be championing the PDEF ensemble forecasting focus following Yuejian Zhu s retirement from the committee. Other Elements Membership Both Richard Swinbank and Yuejian Zhu ended their terms on PDEF leaving 10 PDEF members (John Methven replaced Richard Swinbank as co-chair). We would like to have an additional member with a stronger focus on the use of the TIGGE and S2S ensembles if WWRP can sustain it. Fund raising PDEF helped make the case to EUMETNET (in April 2016) for additional radiosonde launches from the operational network during the NAWDEX campaign. The proposal was successful and funding for 200-250 additional launches was obtained and has been used to good effect to add to the research aircraft flights and ground-based radar and lidar profilers. One example at the end of the campaign involved following the formation of a cut-off cyclone, observed by research flights in the mid-atlantic, as it progressed into southern Europe. Radiosonde launches were activated from ships in the mid-atlantic and in the moist air flow from the Mediterranean ahead of the cut-off, resulting in a high impact precipitation event on the southern Alps. At the same time it was possible to launch additional radiosondes spanning the ridge extending polewards across Iceland into the Arctic (the scientific focus on diabatic effects on the coupled ridge and trough evolution and implications for prediction). Further PDEF initiated, or encouraged, fund raising for research in PDEF relevant areas will be considered in the future. PDEF areas associated with humidity, clouds, precipitation and aerosols need to be advanced because (i) good theory/practice for initializing ensembles of these non-gaussian variables is in its infancy (ii) model errors associated with these processes are likely to be large (iii) these processes limit predictability, and (iv) they are vitally important to coupled model improvement and also to many forms of high impact weather. A possible strategic funding focus on environmental prediction (coupled
modelling with kilometre-scale grids in atmospheric and ocean) is being proposed to the UK funding agency NERC. Meeting Plan The following meetings are relevant to PDEF, with at least one representative attending Date Meeting name Location PDEF participation 19-21 April Hail Workshop Bern Olivia Martius 27 Aug - 1 Sep Symposium on coupled data assimilation and ensemble forecasting Craig Bishop, Oscar Alves 27 Aug - 1 Sep Encourage talks on blocking as part of IAMAS sessions on atmospheric dynamics and predictability 4-6 Sep Third PDEF working group meeting 4-7 Sep SPARC Summer School PPP, S2S meetings and HiWeather, as approporiate John Methven All WG members Olivia Martius, John Methven Annex Distinctions and synergies of DAOS and PDEF interests in Coupled Modeling and Data Assimilation In the Fall of 2015, the DAOS co-chairs (Carla Cardinali, Tom Hamill) and PDEF co-chairs (Craig Bishop, Richard Swinbank) discussed their working groups mutual interests in coupled modelling, ensemble prediction, and data assimilation. A high degree of overlapping interest was found in the areas of ensemble initialization and model-error representation. However, DAOS and PDEF serve differing user groups and hence, it was agreed that across all forecast-lead-times (including those pertaining to coupled modelling) DAOS should address issues of concern to those primarily interested in data assimilation and observing systems while PDEF should address issues of concern to those primarily interested in predictability, dynamics and ensemble forecasting. In the area of coupled modelling, there was general agreement that the ocean-atmosphere coupling and land-atmosphere problem should receive the highest priority initially. Through interaction with the PPP, we also hope to address aspects of ice-ocean and ice-atmosphere coupling. Coupled modelling systems which realistically represent interactions of the atmosphere with land and ocean are important for extended-range prediction, which is particularly relevant to supporting the subseasonal to seasonal prediction (S2S) project. Realistic representation of interactions with the cryosphere is also critical for prediction at high latitudes the focus of the polar prediction project (PPP). Despite strong overlaps in the areas of ensemble forecasting and model error, PDEF will have a much stronger focus on increasing understanding of the dynamics of the coupled system than DAOS, while DAOS will pay more attention to improving observing networks for the coupled system and issues related to the realism of coupled uncertainty estimates in the short-range background forecasts. Weakly coupled DA is the most practical approach to coupled atmosphere-ocean
assimilation in the short term, while the development of (more) strongly coupled DA systems requires a better understanding of coupled uncertainty estimates. We anticipate that experience with coupling the atmosphere with both ocean and land will help with addressing other coupling problems, such as ice-ocean-atmosphere. Both PDEF and DAOS working groups have a common interest in the treatment of model uncertainties, as part of the DA cycling and in ensemble forecast models, respectively. Diagnostic techniques in the context of DA cycling could potentially be very helpful for informing us about how best to represent model uncertainties using stochastic physics methods. Participation of some DAOS members in the proposed joint ECMWF/PDEF workshop on model uncertainty will be very helpful. To help ensure that DAOS and PDEF work synergistically on these issues, it was agreed that a representative from PDEF should be invited to attend future meetings and workshops of the DAOS working group, and vice versa.