The current High Impact Weather forecasting process, whilst varying greatly across the globe, generally involves subjectively interpreting model forecasts of the weather and other data in order to decide whether issuing a warning is appropriate. Model forecasts may be obtained from a very wide range of sources, but typically include deterministic and ensemble forecasts at varying resolutions, supplemented by background observations and ‘environmental’ information (e.g. river levels). This may then be combined with knowledge of the ‘impact response function’, which can be simple and customer-specific (e.g. wind and wave thresholds for ferry operations) or much more complex, as with many public-service warnings. In a number of NMHSs systems are being developed to combine all of these sources of information to produce automated guidance to the forecaster. The forecaster then has a vital role in communicating forecasts and warnings (including the associated uncertainty) in a way that will support decision makers at all levels of their society.
The above process cuts across all the research themes of the HIWeather project. Interpretation of observations and model forecasts is based on understanding of the physical processes involved. As models provide better predictions, including of the impacts, the need for the forecaster to understand the relevant processes becomes wider and more demanding. Currently, human impacts are almost entirely estimated subjectively, but if targeted warnings are to be obtained at high resolution, automated methods will become essential, changing the nature of the warning process. Nevertheless, the forecaster will still need to understand the nature of the impact and society’s vulnerability to it in order to frame warnings and other information appropriately. Advanced techniques of verification can offer information to forecasters for use in interpreting forecasts. However, new communication methods, required to improve the interpretation and use of forecasts and warnings, are likely to have the biggest impact on the forecasting process. The need to promulgate warning information on social media has already produced dramatic changes to the roles of forecasters, and further changes will certainly follow. In some high risk decision making situations, the need to take account of the hazards, the impacted communities / infrastructure and the available mitigation resources has led to the use of collaborative decision making techniques in which diverse experts jointly negotiate a decision. It is likely that use of such approaches will grow, leading to new challenges for NMHSs in managing these organisational, technological and communications interactions. Ultimately it will remain the forecaster’s responsibility to ensure that the information provided is not just useful, but useable and used, requiring that it is delivered in the right form, to the right people, at the right time.
On the global stage there are clear disparities in forecasting and warning capability between developed and poorer nations. Through FDPs and sharing of best practice, this project will foster the dissemination and use of model forecasts and interpretation expertise to address the needs of less developed nations, where commonly the impacts are greatest and resilience least. It can also assist these countries to develop climatologies of weather variables related to impacts, so that frequency of occurrence is understood for use in planning, and to calibrate severe weather products and warnings. In the absence of adequate observational datasets, these may be estimated using reanalyses and hindcasts. Deriving such climatologies from convective-scale models will be a fruitful area of future work.
Work is also needed to more clearly define impact response functions for the application areas of the project – social, economic and environmental – to tie these in with the forecasts and model climate information, and to make such information readily available to forecasters with warning responsibility.
Within WMO, the project will work closely with CBS and the Public Weather Service (PWS) programme to facilitate implementation of the new capabilities developed during the project into operational forecasting (CBS is responsible for operational forecasting and PWS for issuing forecasts to the public). The WMO SWFDP has successfully demonstrated application of the ‘Cascading Forecasting Process’ in which products and new technical capabilities are moved from global to regional and then national centres to strengthen the capacity of NMHSs in developing and least developed countries. The SWFDP has already improved the lead-time and reliability of alerts of high-impact hydro-meteorological events leading to demonstrable protection of life. Close liaison with SWFDP will provide an effective knowledge transfer route for the new capabilities to be developed in HIWeather.
a) Providing the evidence needed by the forecaster to enable effective communication of the hazard situation to key users
b) Providing adequate evidence of track record to enable the forecaster to attach confidence limits to communication
c) Supporting the forecaster in decision making through provision of supporting information to the forecast guidance: latest and recent observations and their match to forecast; agreement between multiple forecast sources; access to historical archives; access to summaries of relevant process studies / training materials; access to scenario assessment tools.
d) Supporting the development of systems to provide automated guidance to the forecaster including identifying research needs across all themes as well as working to quantify the potential and limitations of providing automated guidance for the specific hazards in HIWeather.
a) Led by the Multi-Scale Forecasting theme, raise the level of expertise in high impact weather prediction by involving operational meteorologists in HIWeather research, particularly through evaluation activities in FDPs, testbeds and proving grounds, WMO Training Centres etc.
b) Led by the Predictability & Processes theme, develop the use of models by operational meteorologists to diagnose the origins of hazardous weather features, e.g. using back trajectory techniques. Evaluate the benefit in FDPs, testbeds etc. and publish the results.
c) Led by the Communication theme, in collaboration with operational meteorologists and CBS, develop interpretation aids in high resolution deterministic & ensemble NWP and hazard predictions. Document the material for use in training, e.g. WMO training centres, and as COMET (or similar) training modules.
d) Led by the Evaluation theme, in collaboration with operational meteorologists and CBS, develop verification facilities that enable operational meteorologists to judge the value of new products and capabilities. Evaluate the benefit in FDPs, testbeds etc. and publish the results.
e) Led by the Evaluation theme, in collaboration with operational meteorologists and CBS, develop real-time verification facilities that support operational meteorologists in assessing the accuracy of the current forecast. Evaluate the benefit in FDPs, testbeds etc. and publish the results.
Current observing systems do not meet the time and space scale requirements of high impact weather prediction, nor do they observe most weather impacts nor people’s responses to forecasts and warnings.
All of the research themes have implications for observations. Advancing our understanding requires the collection of highly resolved datasets and their use, with models, to identify the processes that cause high impact weather to develop. Models depend critically on observations for their initialisation. Assessment of human impacts depends on collection of exposure datasets and data on vulnerability. Verification requires data on the key impact variables, while advances in communication methods depend on collection and in-depth analysis of qualitative and quantitative data on people’s responses to different methods.
The current observing networks have largely been developed to meet the requirements of synoptic scale forecasting on a global scale and severe storm nowcasting on a local scale. The global requirement has driven a migration from in situ measurement to satellite-based sounding instruments, while the local requirement has largely been met with increasingly sophisticated radar systems. These remote-sensing systems require supporting in situ data to ensure they remain calibrated. The change of emphasis for local forecasting from forecaster-based nowcasting systems to NWP models is creating a much enlarged requirement for atmospheric monitoring at fine resolution (~10km and less) which may lead to changes to priorities in existing networks, but is unlikely to be fully met by these current approaches to observing. Work is required to:
a) Design observing networks that are fit for purpose on multiple scales
b) Evaluate the future impact of new observations and observing strategies
c) Explore adaptive use of observations
Recent technological developments have raised the possibility of extremely high densities of sensors being deployed, while social networking and crowd sourcing have opened the possibility to obtaining high densities of impact data and potentially of communication, interpretation and uses of warnings, all in real time. However, these opportunities come with enormous challenges in the use of the data, especially in quality control.
a) Design strategies for optimal observation networks for multiple scales (km-scale, mesoscale and synoptic scale), suitable for use in both NWP DA and nowcasting, and deliverable using practicable mixes of observing systems.
b) Account for data assimilation schemes, correlated observation errors, combined sets of diverse observations when designing new observation strategies
c) Identify the satellite sensors (or combinations thereof) that are most relevant for specific hazards.
d) Identify the dependence of hazard prediction improvements on dense observations.
e) Identify observation requirements for monitoring the selected hazards and for assessing forecast accuracy.
f) Quality assurance and control, especially for impact data sources.
g) Data access policies and protocols for data management
a) Led by the Multi-Scale Forecasting theme, develop the use of adjoint-based data impact and/or data denial analysis techniques to km-scale data assimilation experiments so as to establish the value of different data sources in prediction of high impact weather, as measured by metrics developed in the Evaluation theme. Apply these techniques, together with Observing System Experiments (OSEs) and Observing System Simulation Experiments (OSSEs) to assess the costs and benefits of possible future observing system configurations.
b) Led by the Multi-Scale Forecasting theme, demonstrate and evaluate the benefit of enhanced observations, including dense networks of sensors focused on monitoring particular hazards (e.g. temperature for the heat hazard), to the real-time production and communication of hazard warnings in FDPs. Document and publish results, including challenges in gathering, quality controlling and displaying the observations.
c) Led by the Multi-Scale Forecasting theme, in collaboration with NMHSs and spacebased Earth Observation agencies, build on the Committee on Earth Observation Satellites (CEOS) database and other initiatives, to create a catalogue of observations required for monitoring, forecasting, communicating and verifying weather-related hazards and their impacts, of the required spatial and temporal sampling and accuracy, and of candidate new and existing data sources. Promote implementation and real-time international exchange of these observations.
d) Led by the Multi-Scale Forecasting theme and with impacts agencies (emergency management, health etc), build international capability in observing weather-related human impacts and responses for use in monitoring, nowcasting, data assimilation, impact model construction and validation, with particular emphasis on data collection, quality control, and interpretation. The activity will include development of standards for obtaining data from social media, including the use of hashtags (e.g. #snow), considerations of when and how to solicit data, and variations between hazards, countries and cultures.
Most of the relevant impacts are not deterministically predictable on the time and space scales required by users, nor will they ever be due to the fundamental limits on atmospheric predictability. Thus, uncertainty is a common factor in understanding, modelling and communicating high impact weather information.
A fuller appreciation is required of the un-predictability of many severe weather details even at time scales of hours. The ‘deterministic limit’ can be defined as the point in lead time beyond which threshold-based deterministic forecasts are more likely to be wrong than right, i.e. where hits = misses + false alarms, or Critical Success Index (CSI) = 0.5. This is a very useful metric to convey the need to account better for uncertainty. For instance, the deterministic limit is typically only minutes, for convective storms, or hours for many other phenomena. Warnings are needed much further in advance, and so intrinsically have to contain a probabilistic element.
There is strong evidence in the literature of the potential financial benefits of optimal use of probabilistic forecasts / warnings in certain decision scenarios. However it has been difficult to achieve these benefits in practice due to the complexity of most information interpretation and use situations. It is important, therefore, to examine uncertainty communication from the user perspective, and to engage in direct co-education of users and providers. Increased promulgation of probabilistic warnings / forecasts is also needed via experimental or website ‘testbeds’ – and in due course, where appropriate, more promulgation of official warnings in probabilistic terms.
A best practice strategy is needed for progressing from deterministic forecasts to deterministic warnings informed by probabilistic forecasts to probabilistic warnings for appropriate weather and user scenarios. Wherever possible the uncertainty in the weather forecast should be propagated into the impact forecast and should be assessed and communicated to take account of the use to which it will be put by the recipient.
Evaluating more and less successful examples of the use of probabilistic information and the reasons for the successes and challenges, and making use of those lessons, will be important aspects of the mutual education process and the improvement of uncertainty communication.
a) Unresolved processes in NWP models have traditionally been parametrized on the basis that net fluxes are deterministically defined by the grid-scale variables. Recent work in stochastic physics schemes and in model error representation for ensembles has challenged this. We need to gain a better understanding of these results and of their implications for model design.
b) Requirements for hazard advice are frequently presented in deterministic terms because specific decisions have to be made, often quickly. This naturally results in a tendency for the science behind hazard advice to be developed in deterministic terms. We need to establish a culture of thinking from an uncertain framework, of working in probabilistic terms, and of taking risk-based decisions across the breadth of the work of HIWeather, where appropriate from a scientific and user perspective.
a) Led by the Processes and Predictability theme, and in collaboration with mathematics experts, work on stochastic physics and on model error in ensembles will be reviewed and synthesised, and new work promoted, leading to publication of new recommendations for perturbation techniques and parametrization design for ensemble prediction systems.
b) Led by the Communication theme, review and publish the implications of uncertainty in weather forecasts and warnings across the spectrum of the work of HIWeather and how these propagate through the forecast-to-use chain to influence the ability to enhance resilience, with particular emphasis on the creation and communication of probabilistic forecasts and warnings. Promote examples of good practice through interactions with researchers, providers and other stakeholders and through project media opportunities, including a website blog and newsletter.
No single big experimental period is appropriate to the nature of this project. On the other hand, entirely local initiatives are insufficient to advance a global capability. Understanding, modelling and forecasting require high resolution datasets for many types of high impact weather and for the pathways through which the impact is made manifest. Combined field / modelling experiments address this need, focussed on particular weather regimes, preferably on locations and periods when they occur with high likelihood. Research into communication of forecasts, perceptions of recipients, and the actions they take, cannot currently be modelled, so must be undertaken in the field. Given the different responses of different cultures, sampling strategies are critically important. Evaluation depends on enhanced datasets, particularly of the end impact.
These diverse needs can best be met through a planned series of internationally supported experiments incorporating enhanced observations for understanding and forecast development; routine prediction for evaluation and technology transfer; engagement & trialling (with both forecasters and users) for format, reach and relevance, evaluation and trust building. These specialist datasets should be complemented by comprehensive archives of high resolution model outputs over limited areas. It is anticipated that the TIGGE and TIGGE-LAM archives will provide the infrastructure for this.
WWRP has established a set of guidelines for running RDPs and FDPs, including principles of data availability in real time / delayed mode, principles for engaging user communities in the design and execution of FDPs, and principles for performance evaluation. These will form the basis for selection and planning of the cross-cutting experiments in HIWeather. The design of these experiments needs to involve users from the outset so as to ensure that the problem being addressed is aligned with the real problems of those who live and work in the area. Communication, interpretation and the use of information should also be considered from the start to ensure that the users and societal benefit are kept as primary foci.
HIWeather will also participate in testbeds where these address key research issues, and will seek to extend their scope to include prediction of impacts and communication.
Initially, HIWeather will focus on the design and implemention of three experiments: NAWDEX/DOWNSTREAM, LVB-HyNEWS, and ALERT.AR/RELAMPAGO that address key areas of cross-cutting research. Datasets from other recent and planned experiments will be valuable inputs to HIWeather research, and it is likely that HIWeather will wish to become engaged in the design and implementation of additional experiments in future, particularly relating to hazards for which few field experimental data are available.
The THORPEX North Atlantic Waveguide and Downstream Development Experiment (NAWDEX/DOWNSTREAM) is an international initiative to perform coordinated in-situ measurements of disturbances and their evolution along the North Atlantic jet stream and the resulting (high-impact) weather over the USA and Europe. HIWeather links to NAWDEX/DOWNSTREAM will be led by the Predictability and Processes theme and will involve the Multi-Scale Forecasting theme, relating principally to the Urban Flooding and Extreme Local Wind hazards. It is expected that the downstream impact experiments will involve other themes in due course.
This large project is an amalgam of three proposals: A WWRP/WGNR proposal to study weather hazards to fishermen on Lake Victoria, A WCRP/GEWEX proposal to study water balance in the Lake Victoria basin, and an East Africa Commission proposal to study weather impacts on air traffic management. HIWeather is linked mainly with the first of these, specifically to the development of capabilities to monitor and predict hazards related to nocturnal convection that result in many fatalities to fishermen on Lake Victoria. The project will be carried out in association with the East Africa SWFDP. There are links to all HIWeather research themes with particular relevance to Extreme Local Wind hazards, but the lead will be through the Multi-Scale Forecasting theme, and particularly through the WGNR and its successor. For the FDP components, the Communication and Evaluation themes will be key contributors.
The La Plata region of South America is the location for some of the strongest convection in the world, particularly as measured by electrical activity and frequency of flood-generating precipitation. It is also home to a cluster of rapidly developing megacities, including Buenos Aires and Sao Paulo. ALERT.AR is planned to use observations from the RELAMPAGO field experiment to test convective scale NWP models and their coupling to hydrological prediction models, to develop process-based nowcasting techniques for implementation in the NMHSs of the region and to evaluate the impact and communication of forecasts and warnings to decision makers. The links to HIWeather are potentially across all research themes relating to the Urban Flood hazard. A kick-off meeting was held in 2013 and the RELAMPAGO field campaign is planned for late 2017.
Wide gaps in knowledge exist at the present time between the scientific disciplines that must work together to forecast impacts, between research and operations, and between different countries. Separately from work with the external stakeholders, activities will be needed to bridge these gaps if full benefit is to be obtained from the project.
The RDP/FDPs will provide excellent opportunities for bringing together scientists from different disciplines and different countries to address a common problem. Every effort will need to be made to ensure that maximum benefit is obtained from these opportunities, especially for those working in the host country. In addition to planning meetings, it is necessary for this to include working links with local academic institutes and with local emergency response organisations.
Opportunities should also be created to enable sharing of the research results at a higher level through international conferences and/or workshops. These should involve scientists from a broad range of disciplines and countries and should not be split into parallel sessions that separate different research or user communities.
a) The most vulnerable populations and many of the most hazardous events occur in countries that cannot deploy the most advanced hazard forecasting technologies now available. We need to transfer expertise gained in HIWeather to enable research institutes and NMHSs in all countries to effectively contribute to raising their own country’s resilience.
b) Future progress in increasing resilience to weather-related hazards depends on attracting the best researchers to work together in the fields represented in HIWeather. We need to inspire young researchers to work in these fields and give them the inter-disciplinary outlook needed for effective benefits to be gained.
c) Different countries have reached different levels of capability in observing, modelling and forecasting. We need to use different methods of transferring capabilities according to the ability of the recipient NMHS to make use of them.
d) New HIWeather capabilities will be developed on the basis of limited datasets. We need to ensure that they are transferable to other locations and situations before their operational adoption.
e) Language and culture are important factors influencing perception of forecasts and warnings. We need to identify those aspects that must be accommodated to study communication / behaviour and translate knowledge between countries.
a) FDPs will have a primary focus on knowledge transfer to the host NMHS and its local offices and to local academic institutes so as to build up local capability in research and applications. Secondments (in both directions) will be used to support this aim.
b) Summer schools will be held on topics identified in the research pillars to enable young researchers and practitioners, especially from developing countries, to gain knowledge in particular areas in which HIWeather has delivered gains in capability.
c) Reviews of topics identified in the research pillars, will be published as white papers and/or journal issues, enabling widespread access to the results.
d) Results of research and evaluation of its application will communicated through conferences and workshops, including training sessions attached to conferences, and used to create on-line training material for operational meteorologists and users, suitable for use by member countries, e.g. through the COMET programme.
Verification will be necessary to support all themes of the HIWeather project. The Evaluation theme (3.4) identified a number of issues and questions on how to evaluate weather and hazards forecasts and warnings, and societal, economic, and environmental benefits deriving from improved weather and hazard knowledge and communication. Practical applications of verification within each of the HIWeather project themes are discussed below.
The Predictability and Processes theme focuses on understanding the physical processes leading to high impact weather, and therefore requires an evaluation approach tailored to deep understanding. Observational datasets, especially from field campaigns, will be particularly important for describing processes, assimilation into numerical models, and verifying model simulations to establish the validity and credibility of models so that they can be confidently applied in studying the processes of interest. While traditional verification methods have limited usefulness in this context, many of the newer diagnostic approaches may provide useful information to aid understanding of errors in model processes. Errors in model processes can also be investigated through data assimilation, where the relative size of the analysis increments in different variables can provide clues as to which processes are being poorly represented. Advanced visualisation (3D animations, enhanced imagery, etc.) of observation datasets and modelled fields can greatly assist in process understanding and assessing whether the modelled atmospheric flows, evolution of clouds, etc. are well represented.
Verification of multi-scale prediction of weather-related hazards has much in common with routine verification performed at most national meteorological centres, which is used to monitor performance over time, guide development of numerical models, nowcasting systems or other objective guidance products, and assist human forecasters in improving their prediction accuracy and reliability. High impact weather verification should focus on surface variables such as precipitation, wind, temperature, lightning, etc., using both sitespecific and spatial (gridded) approaches to meet the needs of a variety of users.
In recent years there have been guidelines established by WMO discussing best practice verification for deterministic and ensemble NWP, public weather forecasts, precipitation, cloud, and tropical cyclone forecasts, and it is recommended that these guidelines be the starting point for routine verification of high impact weather. Spatial verification and new scores for extremes (EDI, SEDS, etc.) and site-specific verification (e.g., SEEPS) are becoming routinely applied at national centres and should be used in this project. Particular attention should be paid to verifying the timing aspects of weather forecasts and warnings. Real-time verification, even just a picture or a map, would be particularly valuable for forecasters. The HIWeather project should encourage participants to apply best practice verification to experimental forecasts, and it can also collate existing high impact weather verification information from where it is being produced through WGNE, SRNWP, and other international activities.
The meteorological community has less experience in verifying the hazards caused by the weather (floods, landslides, bushfires, etc.). As noted above, observations of hazards are non-standard and difficult to obtain, making routine verification of hazard predictions very difficult. Further, the hazard predictions themselves are often made by agencies outside of the usual meteorological ones. Ensemble prediction, now common in meteorology, may still be quite novel within some hazard communities. The HIWeather project will need to partner with hazard scientists and practitioners who may already be key users of high impact weather information, to assemble forecast and observation datasets and work together to develop appropriate prediction and verification strategies. The meteorological community has a long history of forecast verification know-how which is attractive to those other communities. Some progress in hazard verification has been made, particularly in hydrology (e.g., NOAA’s Ensemble Verification System for streamflow forecasts).
Quantifying the benefit of improvements in high impact weather and hazard prediction on socio-economic impacts is a primary goal of the HIWeather Project. Risk reduction can partly be achieved through more timely and accurate predictions leading to reduced exposure to high impact weather and associated hazards, and facilitating improved preparation and more rapid relief response to reduce negative impacts of high impact weather hazards. Of particular interest will be the added value of probabilistic information which supports more informed decision making on a variety of time scales. The quantitative verification carried out for multi-scale weather and hazard prediction must be propagated through to evaluation of the associated risk reduction. This will involve synthesis with a large variety of demographic, geographic, and other datasets, to enable the exposure and vulnerability components of the risk calculation to be estimated. As with the hazard verification, it will be necessary to partner with scientists and practitioners working in the risk assessment area, and with government agencies holding the relevant datasets (census bureaux, etc.), in order to estimate the risk reduction. Because this is such a vast endeavour, it will be more feasible for the HIWeather project to select some tractable case studies that can be analysed in sufficient depth to allow robust conclusions to be made.
Verifying the benefit of improved communication in achieving more effective response will need to be developed with social scientists in the context of the Communication research theme. Surveys are a common approach to collecting information on the effectiveness of different communication strategies and will be employed here, both to verify that the communication changes have been effective, and to evaluate their impact on the behaviour of the recipients.
a) Understanding the propagation of error and useful information content from meteorological observation, through forecast, to user decisions and outcomes.
b) Connecting verification measures to process understanding and model improvement: both in terms of routine performance and occasions of gross error – forecast busts.
c) Connecting verification measures to the information requirements of users.
a) Led by the Evaluation theme, use reviews, workshops and participation in the design and execution of FDPs, to develop an understanding of the propagation of error and value through the processing chain from meteorological observation and forecast to response and user benefit.
b) Led by the Processes and Predictability theme, develop and apply model diagnostic tools to identify model processes that have caused major forecast errors (busts).
c) Involve users in all stages of the design and execution of FDPs to ensure acceptance of results.
d) Involve HIWeather researchers in Severe Weather Forecast Demonstration Project activities that help NMHSs support each other in providing warnings, including building relationships with national disaster agencies.
The focus on impacts is central to the whole project, with particular input from the Human Impact, Vulnerability & Risk research theme. It will influence the processes studied, the development of models, and the type of communications used.
Impact forecasting requires knowledge of what information is most important to specific audiences for their decisions to reduce impacts and vulnerabilities and mitigate risks. This includes understanding the variables required (e.g., depth of flooding, power outages), the spatial and temporal resolutions (and averaging) required and usable (may differ by audience, even for one type of impact forecast), and the appropriate forms of uncertainty information (e.g., probabilities, scenarios, etc.)
Impacts may be forecast using tools of varying complexity. One of the simplest is to relate the human impact directly to the source of the hazard using an ‘impact response function’. The ability of such simple approaches to provide useful information, both at the awareness raising and warning timescales needs to be established for a varied range of impacts and applications.
Some impact response functions change smoothly with the source while others have discontinuous behaviour. Understanding the differences is important in guiding research in the multi-scale forecasts and processes themes. The dependence of the impact response on regional sensitivities and climate should be emphasised.
a) Stakeholder operational decisions are focused on mitigating impacts that result from specific hazard conditions occurring. Work in all research themes must be focused on what causes or results from the occurrence of these hazard conditions.
a) Led by the Vulnerability & Risk research theme, a catalogue will be prepared of the principal variables that characterise information requirements for stakeholder decision making, including the most significant thresholds, and the nature of the impact response.
b) Led by the Vulnerability & Risk theme, describe an operational forecasting production structure that includes socio-economic impact models and products and promote it through training events, conferences and publications.
A key facilitator of research is the easy availability of field and model data for research purposes. Existing guidelines for the conduct of FDPs & RDPs will go some way to addressing this issue, requiring that as much observational data as possible are made available through the GTS in real time, and that remaining datasets are freely available to researchers within as short a time as possible.
There is no equivalent guideline on the availability of model data at present, but it is proposed that for each RDP/FDP, modelling centres should be encouraged to implement consistently configured km-scale ensemble prediction systems and to make the data available to the TIGGE-LAM archiving centre for as long a period as practicable covering the enhanced observational period. The TIGGE-LAM archiving centre is requested to archive and provide access to these datasets. If it is not possible for TIGGE-LAM to do this, archiving centres should be defined for each activity to take on this role, using an agreed archiving standard.
Standards for storage of social and health survey data should follow best practice in the field, taking account of any precedents in WMO. Appropriate confidentiality and ethical safeguards must be adhered to.
A major source of information for improving global NWP systems has come from historical reanalyses. HIWeather will promote the development and inter-comparison of regional limited-area reanalyses using km-scale data assimilation and modelling systems. The value of such reanalyses depends to a great degree on their accessibility for research. HIWeather will encourage centres that generate such reanalyses to make them freely and easily available for analysis and further processing, using the same archiving standard.
Case studies used in inter-comparisons require participants to have access to data that are not normally shared, such as common definitions of topography and land use, as well as to observations and model fields. Lead centres for such inter-comparisons will ensure that such datasets are easily accessible in standard formats.
a) Archiving, research access and (where applicable) public access to high resolution forecasts, observations and products from HIWeather field campaigns, case studies, inter-comparisons, and surveys in standard formats.
a) Led by the Multi-Scale Forecasting theme, agree data format, storage & access standards for the project, taking account of existing standards and WMO guidance.
b) Led by the Multi-Scale Forecasting theme, ensure that as much data as possible is shared in real time during field experiments.
c) Led by the Vulnerability and Risk theme, ensure that necessary ethical safeguards are built into all survey work undertaken.
d) Led by the Multi-Scale Forecasting theme, agree data archiving and access structures for each data-generating component of the project prior to start of data production. e) Led by the Multi-Scale Forecasting theme, ensure that metadata and data summaries are published as appropriate.