Development and evaluation of the world’s first multi-model near-term prediction system.

Initialised multi-year predictions have evolved to the point where semi-operational predictions are produced annually from ten global modelling centers. This represents both an opportunity and a challenge for decision-makers in how they make use of this information. New forecast verification tools and establishment of common case studies provide a common basis to evaluate the skills and relative strengths of the different prediction systems. One of the ground breaking objectives of EUCP is to explore whether a new multi-model/multi-operational center near-term prediction system is able to provide more skillful information over individual single model predictions (WP1). The multi-model ensemble is less sensitive to individual model structural errors and is, therefore, able to better capture climate variations. Evaluation of the new multi-model near-term prediction system will draw heavily on the new forecast verification tools developed in WP1. This evaluation of where this new system has skill over individual operational systems will inform the evaluation and data synthesis (WP5) and the nature of the initialised forecasts presented to end users.

Including a much wider climate model-observational evaluation targeting specific processes that have been shown to link to the plausibility of future changes.

Many of the high profile climate science advances in the past five or so years have been in the development of “emergent constraints”. Larger collections of climate simulations (CMIP3, CMIP5) that explore a greater diversity of historical and future pathways have enabled relationship between historical climate model biases and the magnitude of future changes to be identified. These relationships combined with good observations point to the potential for using targeted climate model-observed comparisons to narrow the range of projected climate changes that can be considered plausible. To date information from the growing collection of such “emergent constraints” has not yet fed through into National Climate Projections. However, at the same time very little work has been done to confront climate model projections with multiple emergent constraints or explore how well, if at all, these combine. EUCP will first provide a synthesis of the current global status of emergent constraints and then explore the impact of each on the range of future changes. The challenge in doing so is how to make sense of a wider range of different constraint information, some of which will be complementary and some contradictory (that may arise due to current structural model errors for example). By applying these constraints consistently (WP2), EUCP will provide a coherent picture of how information from these different sources will be combined. This will ensure that EUCP projections will be able to down-weight less credible projections but the use of multiple observational metrics will also avoid overly tight constraints that can arise from single metrics due to structural model errors. This will be a world first and will offer insights that will inform the wider scientific understanding of the information and value of current “emergent constraints” for future climate projections. Assessing the various projection data against these multiple observational comparisons will represent a key basis of confidence that will inform their use in risk-based decision making.

The impact of explicitly resolving convective-scale processes in regional climate modelling.

Many of the uncertainties in modelling the climate response are due to the need to parameterise processes that occur on smaller scales than the models are able to explicitly resolve. As advances in computing enable higher resolutions, processes that previously were parameterised can be resolved explicitly. Regional climate models are approaching a resolution barrier, with the release of simulations that can, for the first time, be considered “convection permitting” (resolving many but not all of the convective scale process) (Prein et al., 2015). Early results suggest that while such models reproduce similar pictures of mean projected rainfall changes, they suggest a shift in when rainfall occurs to less frequent but heavier rainfall events (Kendon, 2017). Any change in the intensity of heavy rainfall events can be expected to impact a number of sectors, but the small number of existing studies make it difficult to infer robust messages on what this new modelling capability will bring. EUCP will coordinate a set of new regional climate model simulations, at the convective permitting scale (1-3 km, both with a pan European domain and with more targeted modelling of the Alpine region) across models from nine contributing regional modelling centres. This will enable a first systematic assessment of what new insights these convection permitting models bring and what (and where) are the robust messages for climate change projections. In doing so it will address important open questions on the implications for climate impact that early model studies have raised.

EUCP draws on data from a very broad range of climate simulations and develops new and novel climate model-observational benchmarks and constraints, verification tools and comparisons and makes use of existing observational evaluations. An overarching objective of EUCP is to synthesize these diverse strands of evidence to support each of the climate products (from probabilistic near-term predictions and longer-term projections, through ensembles of climate simulations, to storyline scenarios used as entry points for end users). Whilst many of the underpinning climate data sources are not unique to EUCP, it is the coherent evaluation of the consistency of these multiple lines of evidence that represents the novel outcome.

Amongst the multiple lines of evidence, EUCP will include:

  • Consistency of future changes. Do projections/predictions agree on the sign and magnitude of future changes? Do different climate projection approaches lead to agreement? Do the new convection permitting models agree or diverge from the existing resolution climate projections, and if so for what temporal and spatial scales and locations and what variables?
  • Evaluation of the climate model skill at reproducing a broad range of observed metrics (going beyond the current evaluation of climatological fields). This includes new near-term forecast verification benchmarks, new observational insights that draw on process led assessments and the field of “emerging constraints” in longer term projections); Consistency with observed trends (e.g., are projected increases in high intensity rainfall evident in tendencies found in the observations?); and constraints on the forced response arising from detection and attribution results.
  • Assessment of how individual storyline scenarios, or plausible future realizations, sit within the wider range of projected climate changes. This includes exploration of where, and for what variables and scales should we expect the signal to emerge from natural climate variability? Where does a particular future realization sit within the context of wider estimates of future climate risk?
  • Assessments of the underlying physical mechanisms that drive these changes and how consistent these are to our physical understanding of these processes in the real world. For example, do the models adequately resolve or represent the processes that are known to affect a particular impact? Is there added value in addressing certain high impact events in models that meet certain criteria (e.g. use of new non-hydrostatic model simulations to look at heavy precipitation; “high top” model to capture blocking) and how can we best make use of this added value?

In some cases, the proposed analysis can be envisaged to lead to narrower, more tightly constrained climate predictions/projections. However, the intended outcome of this process is to provide a better-quantified estimate of the risk and a greater confidence base to the users. Information on the evidence base (types of sources, consistency, quality) and the level of agreement provides the important context that helps inform how and where this information can be used.

Co-development of Climate Projections

In many of the existing National Projections, the science drivers have often shaped the form that projections have been delivered in (e.g., the need to incorporate new sources of uncertainties; new modelling capability and analysis approaches). Users of the projection data bring their own drivers and expectations (needs for accessibility, data to inform risk assessments or to drive impacts models, or provide a top-level narrative). There is a growing recognition of benefits and the need to co-develop climate projections, iterating between science and user bases. This act of co-development enables the realization of climate products that meets user’s needs whilst being consistent with best available information from science drivers. The recent KNMI’14 illustrates how a co-developed set of Climate Projections can change the nature of climate products that are delivered. The various sources of uncertainties are not transparent in the climate narratives produced. Instead, they represent storylines that span the underlying climate and socio-economic uncertainty. By providing a limited number of narratives they lowered the barrier of entry for end users (who, for example, do not require understanding their own climate vulnerabilities prior to using the projections). EUCP establishes targeted engagement with super-users (engaged end users of climate data who are already aware of some of their own climate exposure) in a series of demonstration studies (WP4). These studies are city-scale urban flooding; coastal erosion and flooding; how information on changes in hydroclimatic extremes can be effectively communicated; and working with the wind energy sector to explore climate drivers of wind stilling trends and calming events. In each case, the project (via WP4 and WP6) establishes two-way interactions with WP1-3 in identifying and developing the information which will inform their decisions. These demonstration studies will occur within the context of a wider end-user engagement. WP6 will feed developments in climate prediction and projection data into wider discussions about bridging the gaps between providers and users. Through the Multi-user Forums (WP6), EUCP will bring together representatives from user organisations drawn from financial organisations, public bodies and authorities, civil society organisations, businesses, risk-related partnerships, and academic institutions, to reflect on the scientific, technical and economic questions underpinning the EUCP objectives; and to review, discuss, co-design and disseminate the preliminary and final results. This framework will play a crucial role co-developing the EUCP innovation and will stimulate collaboration and co-production of knowledge among the end-users, purveyors and climate service providers. The Platform will comprise user organizations and will work through focussed teams addressing individual sectors and will provide the overview of EUCP development via a series of annual conferences that will inform the science directions in WPs 1-3.

Provision of a baseline of climate information that will facilitate downstream adaptation

Beyond the targeted user engagement, EUCP intends to build on existing EC-funded projects, including those under Horizon 2020 SC5 and Blue Growth. Central to this will be the interactions with Climateurope, as outlined in WP6. This will include working with the Blue-Action project to support emergent businesses in rethinking/redesigning their business models in light of climate data that they receive through this project.

EUCP will demonstrate the delivery of a pan-European climate prediction system and is not intended to address all user needs. However, outside of the planned collaboration, the release and availability of data will provide a new common baseline of climate data across Europe, that is consistent, coherent and represents the best available scientific data to date. The impact of this is expected to be most obvious for those European countries where National Climate Projection data does not currently exist. In doing so, it removes one of the many barriers to informed climate adaptation decision making. The knowledge and data portal will also provide the guidance documentation in the information and use of EUCP data, as well as documenting the case studies that exploit this data (WP4). In doing so, it should raise the accessibility to current information on how climate variability and climate change will affect Europe as a whole and remove much of the current geographical inequality that exists in climate data access.