Seminars (CLP)

A rather prevelant picture of the development of climate models throughout the 20th Century, is for the idealised, simplified, and hence mathematically tractable models of climate to be the focus of mathematicians, leaving to engineers, the "brute force" approach of developing ab initio Earth System Models. I think we should leave this paradigm in the 20th Century, where it belongs: for one thing, the threat of climate change is too important and the problems of predicting climate reliably too great. For the 21st Century, I propose that mathematicians need to engage on innovative methods to represent the unresolved and poorly resolved scales in ab initio models, based on nonlinear stochastic-dynamic methods. The reasons are (at least) threefold. Firstly, climate model biases are still substantial, and may well be systemically related to the use of deterministic bulk-formula closure - this is an area where a much better basic understanding is needed. Secondly, deterministically formulated climate models are incapable of predicting the uncertainty in their predictions; and yet this is a crucially important prognostic variable for societal applications. Stochastic-dynamic closures can in principle provide this. Finally, the need to maintain worldwide a pool of quasi-independent deterministic models purely in order to have an ad hoc multi-model estimate of uncertainty, does not make efficient use of the limited human and computer resources available worldwide for climate model developement. The development of skilful stochastic-dynamic closures will undermine the need for such inefficient use of human resources. As such, a very grand challenge for the science of climate prediction is presented in the form of a plea for the engagement of mathematicians in the development of a prototype Probabilistic Earth-System Model. It is hoped that this Newton Institute Programme will be seen as pivotal for such development.

Striking developments in the climate system in recent years have reinforced the view that anthropogenic radiative forcing is unlikely to cause a smooth transition into the future. Drought in the Amazon in 2005, record Arctic sea-ice decline in 2007, accelerating loss of water from the Greenland and West Antarctic ice sheets, and an extraordinary Asian summer monsoon in 2010, have all made the headlines. These large-scale components of the Earth system are among those that we have identified as potential ‘tipping elements’ – climate sub-systems that could exhibit a ‘tipping point’ where a small change in forcing causes a qualitative change in their future state. The resulting transition may be either abrupt or irreversible, or in the worst cases, both. In IPCC terms such changes are referred to as “large-scale discontinuities”. Should they occur, they would surely qualify as dangerous climate changes. Recent assessments suggest that the traditional view of the likelihood of tipping points as very low probability events should be revised upwards - especially if we continue business-as-usual. Given this, is there any prospect for providing societies with a useful early warning signal of an approaching climate tipping point? The talk will have two main aims. Firstly, we want to review (and slightly revise) the list of potential tipping elements, providing some updates, especially where there is new insight into the mechanisms behind them, or new information about the proximity of tipping points. Secondly, we want to present our ongoing work to try and develop robust methods of identifying and anticipating tipping points (in particular, bifurcations) in the climate system. Our latest application of these methods to sea surface temperature data suggest that a new climate state may be in the process of appearing, particularly in the Arctic and northernmost Atlantic region.

The physical basis for current methods of stochastic parametrization in NWP/climate models is reviewed and their plausibility assessed with respect to unresolved or near-gridscale meteorological phenomena. This issue is closely related to that of the predictability of convective scale and mesoscale weather systems. The coarse-graining strategy is described and applied to high-resolution NWP model forecast output and cloud-resolving model simulations of deep, tropical convection. The results are used to provide some constraints on the stochastic backscatter and the perturbed physical tendency approaches.

The climate system has a wide range of temporal and spatial scales for important physical processes. Examples include convective activity with an hourly time scale, organized synoptic scale weather systems on a daily time scale, extra-tropical low-frequency variability on a time scale of 10 days to months, to decadal time scales of the coupled atmosphere-ocean system. An understanding of the processes acting on different spatial and temporal scales is important since all these processes interact with each other due to the nonlinearities in the governing equations. Most of the current problems in understanding and predicting the climate system stem from the multi-scale nature of the climate system in that all of the above processes interact with each other and the neglect and/or misrepresentation of some of the processes lead to systematic biases of the resolved processes and uncertainties in the climate response. A better understanding of the multi-scale nature of the climate system will be crucial in making more accurate and reliable weather and climate predictions. In my presentation I will discuss systematic strategies to derive stochastic models for climate prediction. The stochastic mode reduction strategy accounts systematically for the effect of the unresolved degrees of freedom and predicts the functional form of the effective reduced equations. These procedures extend beyond simple Langevin equations with additive noise by predicting nonlinear effective equations with both additive and multiplicative (state-dependent) noises. The stochastic mode reduction strategy predicts rigorously closed form stochastic models for the slow variables in the limit of infinite separation of time-scales.

The simplest class of stochastic models relevant to geophysical applications consist of a linearization of the dynamical system and the addition of constant multivariate stochastic forcing. Such stochastic systems are known as finite dimensional Ornstein Uhlenbeck systems and have wide application. In this talk we describe a general decomposition of the equilibrium spectrum of such processes. This is of interest in applications since spectra of long time series are commonly robustly defined from observations. We apply this formalism to the case of ENSO where it is often argued that there is a dominant normal mode. Here we argue that the decadal part of the ENSO spectrum can be simply explained by the stimulation of the cross spectrum of the dominant normal mode. The cross spectrum is dependent on the ENSO cycle phase meaning that this mechanism implies that the different ENSO phases have different spectral strengths at decadal frequencies.

This paper discusses rate-dependent tipping points related to a novel excitability type where a (globally) stable equilibrium exists for all different fixed settings of a system's parameter but catastrophic excitable bursts appear when the parameter is increased slowly, or ramped, from one setting to another. Such excitable systems form a singularly perturbed problem with at least two slow variables, and we focus on the case with locally folded critical manifold. Our analysis based on desingularisation relates the rate-dependent tipping point to a canard trajectory through a folded saddle and gives the general equation for the critical rate of ramping. The general analysis is motivated by a need to understand the response of peatlands to global warming. It is estimated that peatland soils contain 400 to 1000 billion tonnes of carbon, which is of the same order of magnitude of the carbon content of the atmosphere. Recent work suggests that biochemical heat release could destabilize peatland above some critical rate of global warming, leading to a catastrophic release of soil carbon into the atmosphere termed the ``compost bomb instability''. This instability is identified as a rate-dependent tipping point in the response of the climate system to anthropogenic forcing (atmospheric temperature ramping).

The ensemble approach has originally been derived in probabilistic medium-range weather forecasting, and is now broadly used in numerical weather prediction, seasonal forecasting and climate research on a wide range of time scales. Applications geared towards climate projections are usually based on a heterogeneous ensemble with typically a mere handful of ensemble members, stemming from different models in an only partly coordinated framework. An important feature of ensemble approaches in climate research is the inability to rigorously quantify climate model biases. While biases of climate models are monitored for the control period, the lack of long-term comprehensive observations (on the centennial time-scales considered) implies that it is difficult to decide how the model biases will change with the climate state. In contrast to other studies, we look not only at 20 or 30 year averages, but also at the interannual variability. This allows us to consider additive and multiplicative biases. In the talk, I will discuss two plausible assumptions about the extrapolation of additive biases, referred to as the ``constant bias'' and ``constant relation'' assumptions. The former is used implicitly in most studies of climate change. The latter asserts that over-/underestimation of the interannual variability in the control period leads also to over-/underestimation of climate change, and this assumption is closely related to the statistical post-processing of seasonal climate predictions. In addition we explicitly allow the additive and multiplicative model biases to change between control and scenario periods, resolving the resulting lack of identifiability by the use of informative priors. An analysis of of GCM/RCM simulations from the ENSEMBLES project shows that bias assumptions critically affect the results for several regions and seasons.

Throughout Earth’s history, life has increased greatly in abundance, complexity, and diversity. At the same time, it has substantially altered the Earth’s environment, evolving some of its variables to states further and further away from thermodynamic equilibrium. For instance, concentrations in atmospheric oxygen have increased throughout Earth's history, resulting in an increased chemical disequilibrium in the atmosphere as well as an increased redox gradient between the atmosphere and the Earth's reducing crust. These trends seem to contradict the second law of thermodynamics, which states for isolated systems that gradients and free energy are dissipated over time, resulting in a state of thermodynamic equilibrium. This seeming contradiction is resolved by considering planet Earth as a coupled, hierarchical and evolving non-equilibrium thermodynamic system that has been substantially altered by the input of free energy generated by photosynthetic life. Here, I present this hierarchical thermodynamic theory of the Earth system. I first present simple considerations to show that thermodynamic variables are driven away from a state of thermodynamic equilibrium by the transfer of power from some other process and that the resulting state of disequilibrium reflects the past net work done on the variable. This is applied to the processes of planet Earth to characterize the generation and transfer of free energy and its dissipation, from radiative gradients to temperature and chemical potential gradients that result in chemical, kinetic, and potential free energy and associated dynamics of the climate system and geochemical cycles. The maximization of power transfer among the processes within this hierarchy is closely related to the proposed principle of Maximum Entropy Production (MEP). The role of life is then discussed as a photochemical process that generates substantial amounts of additional free energy which essentially skips the limitations and inefficiencies associated with the trans fer of power within the thermodynamic hierarchy of the planet. In summary, this perspective allows us to view life as being the means to transform many aspects of planet Earth to states even further away from thermodynamic equilibrium than is possible by purely abiotic means. In this perspective pockets of low-entropy life emerge from the overall trend of the Earth system to increase the entropy of the universe at the fastest possible rate. The implications of the theory presented here are discussed regarding fundamental deficiencies in Earth system modeling, applications of the theory to reconstructions of Earth system history, to evaluate human impacts and regarding the limits of renewable sources of free energy for future human energy demands.

Most investigations of the MEP hypothesis have used climate models which do not explicitly simulate the physics and dynamics of the climate system (such as Paltridge's model), or in which they are radically simplified (such as a dry GCM). We have instead concentrated on entropy analysis of the HadCM3 atmosphere-ocean general circulation model (AOGCM), the kind of model used for prediction of 21st-century global climate change. In the AOGCM, we diagnose the entropy sources and sinks directly from the diabatic heating terms. The rate of material entropy production of the climate system (i.e. not including thermal equilibration of radiation) is about 50 mW m-2 K-1. The largest part of the material EP (about 38 mW m-2 K-1), is due to sensible and latent heat transport. When we vary parameters in the physical formulation of the AOGCM, MEP might suggest that the most realistic version is the one with the largest EP. However, in the AOGCM there is no maximum in EP, for two reasons. First, the strongest influence on EP is the throughput of energy from the net shortwave absorption, which is very sensitive to model parametrisation, rather than the anticorrelation of heat flux and temperature gradient seen in simple models when net shortwave absorption is fixed. This dependence comes particularly from the dominance of EP by the hydrological cycle, which intensifies monotonically with the global average temperature. Second, the EP predominantly comes from vertical heat transport, and to achieve a maximum with fixed shortwave heating implies an unrealistic vertical temperature gradient and/or unphysical longwave emissivity. There is, however, a maximum in KE dissipation in the atmosphere, similar to Lorenz's (1960) conjecture, associated with a smaller part of the material EP (about 13 mW m-2 K-1).

The Integrated Forecasting System (IFS) is a sophisticated software system for weather forecasting, which was jointly developed by the European Centre for Medium-Range Weather Forecasts (ECMWF) and Meteo France. All applications needed for generating operational weather forecasts are included, such as data assimilation, atmospheric model and post processing. The IFS is used for deterministic 10 day forecasts and ensemble forecasts with forecast ranges from 15 days for the medium range EPS, 32 days for the monthly forecast up to 13 month for the seasonal forecasts. In addition to a good deterministic forecast model as basis of the ensemble prediction system, the ingredients needed to produce good ensemble forecasts are realistic and appropriate representations of the initial and model uncertainties. The stochastic schemes used for the model error representation will be presented. These are the Spectral Stochastic Backscatter Scheme (SPBS) and the Stochastically Perturbed Parametrization Tendency Scheme (SPPT). The basis of both schemes is a random spectral pattern generator, in which the spectral coefficients are evolved with a first order auto-regressive process. The resulting pattern varies smoothly in space and time with easy to control spatial and temporal correlation. The two schemes address different aspects of model error. SPPT addresses uncertainty in existing parametrization schemes, as for example parameter settings, and therefore generalizes the output of existing parametrizations as probability distributions. SPBS on the other hand describes upscale energy transfer related to spurious numerical dissipation as well as the upscale energy transfer from unbalanced motions associated with convection and gravity waves, process missing in conventional parametrization schemes. Cellular Automata (CA) are an alternative way for generating random patterns with temporal and spatial correlations. A pattern generator based on a probabilistic CA was implemented in the IFS. The implementation allows the interaction of model fields with the CA, i.e. the characteristics of the CA are influenced by the atmospheric state. The impact of the stochastic schemes on the forecast skill will be presented for different forecast ranges.

Our understanding of the climate system has been revolutionized by the development of sophisticated computer models. Yet, these models are not perfect representations of reality, because they remove from explicit consideration many physical processes which are known to be key aspects of the climate system, but which are too small or fast to be modelled. Examples of such processes include gravity waves. This talk will give several examples of implementations and impacts of stochastic sub-grid representations in atmosphere and ocean models.

The climate belongs to the class of non-equilibrium forced and dissipative systems, for which most results of quasi-equilibrium statistical mechanics, including the fluctuation-dissipation theorem, do not apply. We show for the first time how the Ruelle linear response theory, developed for studying rigorously the impact of perturbations on general observables of non-equilibrium statistical mechanical systems, can be applied to analyze the climatic response. We choose as test bed the Lorenz 96 model, which has a well-recognized prototypical value. We recapitulate the main aspects of the response theory and propose some new results. We then analyze the frequency dependence of the response of both local and global observables to perturbations with localized as well as global spatial patterns. We derive analytically the asymptotic behaviour, validity of Kramers-Kronig relations, and sum rules for the susceptibilities, and related them to parameters describing the unperturbed properties of the system. We verify the theoretical predictions from the outputs of the simulations with great precision. The theory is used to explain differences in the response of local and global observables, in defining the intensive properties of the system and in generalizing the concept of climate sensitivity to all time scales. We also show how to reconstruct the linear Green function, which maps perturbations of general time patterns into changes in the expectation value of the considered observable. Finally, we propose a general methodology to study Climate Change problems by resorting to few, well selected simulations and discuss the specific case of surface temperature response to changes of the $CO_2$ concentration. This approach may provide a radically new perspective to study rigorously the problem of climate sensitivity and climate change.

This work aims at improved computation of covariances and of multiscale structures such as clouds, in the Weather Research and Forecasting (WRF) data-assimilation (WRFDA) system, in particular the horizontal factor of the control-variable transform used to optimize the forecast initialization. Better representation can be achieved in the horizontal transform by wavelet-compression techniques that have been proven in many other applications.

In this work, two past obstacles to effective incorporation of wavelets in limited-area models such as WRF are resolved: isometric-injective (i.e., energy preserving, left-invertible) wavelets avoid boundary-condition assumptions at any scale; and these wavelets can be applied to non-dyadic data lengths. A summary technical description of these improved wavelets and their implementation into WRFDA is presented. By retaining only a diagonal background-covariance matrix in wavelet space, appropriate heterogeneity is obtained for the model-space covariances.

A second wavelet application is to partition observation error into a part due to poor representation (e.g., too-coarse resolution), and a residual, using a novel criterion in wavelet space. Other methods to construct inhomogeneous anisotropic covariance models are cited, and other potential technical improvements are discussed.

Data assimilation aims to solve one of the fundamental problems of numerical weather prediction - estimating the optimal state of the atmosphere given a numerical model of the dynamics, and sparse, noisy observations of the system. A standard tool in attacking this filtering problem is the Kalman filter.

We consider the problem when only partial observations are available. In particular we consider the situation where the observational space consists of variables which are directly observable with known observational error, and of variables of which only their climatic variance and mean are given. We derive the corresponding Kalman filter in a variational setting.

We analyze the variance constraining Kalman filter (VCKF) filter for a simple linear toy model and determine its range of optimal performance. We explore the variance constraining Kalman filter in an ensemble transform setting for the Lorenz-96 system, and show that incorporating the information on the variance on some un-observable variables can improve the skill and also increase the stability of the data assimilation procedure.

Using methods from dynamical systems theory we then investigate systems where the un-observed variables evolve deterministically (but chaotically) on a fast time scale.

This is joint work with Lewis Mitchell and Sebastian Reich.

This talk presents challenges and first results on how to do stochastic parametrizations for deterministic systems with some underlying conservation laws such as energy conservation. We introduce the problem, and then consider a time scale separated Hamiltonian toy model where the fast chaotic dynamics will be modelled by noise in such a way as to preserve the energy conservation of the full deterministic system. The reduction is done in two stages: First projecting the noise onto the energy manifold, and then subsequently performing singular perturbation theory.

This is joint work with David Lewis.

The Northwest Atlantic has the largest annually averaged oceanic surface heat loss in the world's ocean, with a wintertime spatial average of over 200 W/m2. In addition, it is at this location in the Northern Hemisphere that the zonally averaged meridional heat transport by the ocean reduces abruptly to below 0.5 PetaWatts with the atmosphere taking over the bulk of heat transport. Interannual changes of the exchange of heat between the ocean and the atmosphere are large in this region with magnitudes exceeding 50 W/m2 or more averaged over the Gulf Stream, equivalent about to 0.1 PetaWatts. To quantify the source of interannual changes in surface heat flux and heat storage, we examine the ocean heat budget in four different ocean models, a 1/10-degree prognostic ocean model (POP), a 1/3-degree assimilative model of the North Atlantic (Mercator), a diagnostic model, and a 18km degree Green;s function assimilation model (ECCO2). The last three models solutions all depend directly on the observed altimetric sea surface height. We focus the Gulf Stream once it leaves the coast. While the models disagree in detail, they do agree on some points. The ocean can locally store substantial quantities of heat on theses times scales. The changes in heat content on interannual time scales are concentrated in the upper 400 m of the water column while for monthly times scales, they are concentrated in the upper 100 m. Monthly heat content changes are controlled by the atmosphere with the ocean lagging the surface heat flux by one month. On interannual times scales, heat content changes are controlled by horizontal heat transport convergence with only a secondary role for the surface heat fluxes. At the same time, the surface fluxes are negatively correlated with the oceanic heat content, lagging it by 3 months. This indicates that the surface fluxes damp the heat content changes in the ocean instead of forcing them, suggesting an active role of ocean dynamics in the climate system. Additional comments on the use of ocean models for studies of the ocean's role in climate are also given.

Bayesian models and methods have arguably been the gold standard for probabilistic assessment of scientific uncertainty for 250 years. The theory of belief functions, also known as Dempster-Shafer (DS) theory referring to its founds in the 1960s and 1970s, has the potential to break through limitations and confusions that are the legacy of the major conceptual and technical advances of the 20th Century. I start with a brief sketch of sources, motivations, and key elements of DS, including why this may be a propitiuous time for a new look at a not-so-new formalism. For statisticians, key ideas are that probabilities are personal and extend beyond the Bayesian paradigm to include probabilities of "don't know," providing new tools for addressing problems of robustification, multiparameter estimation, and nonidentifiability. For climateology, DS opens new ways to think about what is predictable and what is not predictable.

Chair: Jonathan Leake (Science & Environment Editor, The Sunday Times). Panel: Tim Lenton (Earth System Modelling, University of East Anglia); Tim Palmer; Vicky Pope (Head of Climate Change Advice, the Met Office); Alan Thorpe (Chief Executive, Natural Environment Research Council).

It has never been a more exciting time to be a climate modeler. International coordination of model experiments by the Coupled Model Intercomparison Project (CMIP5) will see the estimated production of over 2 petabytes of model output from 20 modeling centers and 40 different climate models (downloading the data on a home broadband would take 25 years). There will be new model functionality in terms of the processes represented in the models including chemistry and biology, new forcing scenarios including palaeoclimate and idealized cases and new experiments initialized with observations to look at near-term climate variability and change. Moreover there is an unprecedented interest in, and scrutiny of, climate model projections from fellow scientists, from the public and from governments.

How on earth are we to make sense of this information overload? This talk will review some of the approaches that we expect will be used to analyses the new CMIP5 multi-model database. Some approaches rely on physical understanding of the climate system to make sense of the data. Some use simple statistical approaches to rationalize the output. In some specific cases, more complex statistical approaches may be applied carefully. Finally, all the approaches will have to be synthesized to provide a summary of the state of climate modeling science. This challenge will be discussed.

Climate changes at time scales of several (tens) of thousands years like glacial interglacial cycles may be viewed as an 'emergent feature' of the climate system. They can be understood at different levels : from a general circulations prospective (how changes in astronomical elements affect the hydrological, nutrient and carbon cycles); from a dynamical system prospective (locate and characterize bifuraction points, detect synchronisation phenemena, identify couplings between different time scales... ); or from a statistical prospective (estimate the probability of events and assess the predictability of the climate system at these time scales). The ambition of a general theory Pleistocene climate is to merge these approaches.

The recent mathematical developments reviewed during the present Newton Institute constitute promising avenues to this end. For example, statistical emulators allow to explore in depth the input and parameter spaces of general circulation simulators, including their sensitivity to the astronomical forcing. Monte-Carlo statistical methods allow to calibrate low-order stochastic dynamical systems, and guide the process of criticism and selection of models. The purpose of this talk is to summarise advances gained during the Newton Institute along these lines.

Although modern climate simulators represent our best available understanding of the climate system, projections can vary appreciably between them. Increasingly therefore, users of climate projections are advised to consider information from an "ensemble" of different simulators or "multimodel ensemble" (MME).

When analysing a MME the simplest approach is to average each quantity of interest over all simulators, possibly weighting each simulator according to some measure of "quality". This approach has two drawbacks. Firstly, it is heuristic: results can differ between weighting schemes, leaving users little better off than before. Secondly, no simulator is uniformly better than all others: if projections of several different quantities are required the rankings of the simulators (and hence the implied weights) may differ considerably between quantities of interest.

A more sophisticated approach is to use modern statistical techniques to derive probability density functions (pdfs) for the quantities of interest. However, no systematic attempt has yet been made to sample the range of possible modelling decisions in building a MME: therefore it is not clear to what extent the resulting "probabilities" are in any way relevant to the downstream user.

This talk presents a statistical framework that addresses all of these issues, building on Leith and Chandler (2010). The emphasis is on conceptual aspects, although the framework has been applied in practice elsewhere. A mathematical analysis of the framework shows that:

(a) Information from individual simulators is automatically weighted, alongside that from historical observations and from prior knowledge. (b) The weights reflect the relative value of different information sources for each quantity of interest. Thus each simulator is rewarded for its strengths, whereas its weaknesses are discounted. (c) The weights for an individual simulator depend on its internal variability, its expected consensus with other simulators, the internal variability of the real climate, and the propensity of simulators collectively to deviate from the real climate. (d) Some subjective judgements are inevitable.

Reference: Leith, N.A. and Chandler, R.E. (2010). A framework for interpreting climate model outputs. J. R. Statist. Soc. C 59(2): 279-296.

This talk will summarise progress on the application of three strands of the Newton Institute Mathematical and Statistical Approaches to Climate Modelling and Prediction Programme, which are all connected through the general notion of turning noise into signal:

(a) Maximum Entropy Production (MEP) principles of the climate system – application to surface-to-atmosphere turbulent fluxes. (b) Time-series techniques to give forewarning of climate tipping points – application to possible climate-driven Amazon forest dieback. (c) Fluctuation theories to determine climate system sensitivities – using interannual variability of CO2 as a constraint on the sensitivity of tropical land carbon to climate change.

In the case of (b) and (c) short-timescale variability is used to determine the vulnerability of the system to an abrupt change and a warming-induced release of land carbon, respectively. In (a), the application of MEP to vertical turbulent fluxes puts these “noisy” fluxes at centre-stage in the climate system, rather than being treated as a residual term in the surface energy balance.

The Newton Institute programme has stimulated progress in all three of these areas. This talk will summarize the current state-of-play and suggest priorities for follow-on research.

Successful communication of knowledge to climate change policy makers requires the careful integration of scientific knowledge in an integrated assessment that can be clearly communicated to stakeholders, and which encapsulates the uncertainties in the analysis and conveys the need for using a risk assessment approach. It is important that (i) the system is co-designed with the users (ii) relevant disciplines are included (iii) assumptions made are clear (iv) the robustness of outputs to uncertainties is demonstrated (v) the system is flexible so that it can keep up with changing stakeholder needs and (vi) the results are communicated clearly and are readily accessible. The "Community Integrated Assessment System" (CIAS) is a unique multi-institutional, modular, and flexible integrated assessment system for modeling climate change which fulfils the above criteria. It differs from other integrated models in being flexible, allowing various combinations of component modules, to be connected together into alternative integrated assessment models. Modules may be written at different institutions in different computer languages and/or based on different operating systems. Scientists are able determine which particular CIAS coupled model they wish to use through a web portal. This includes the facility to implement Latin hypercube experimental design facilitating formal uncertainty analysis. Further exploration of robustness is possible through the ability to select, for example, alternative climate models to address the same questions. It has been applied to study climate change mitigation, through for example the AVOIDing dangerous climate change project for the UK Department of Energy and Climate Change (DECC), in which the avoided impacts (benefits) of alternative climate policies were compared to no-policy baselines. AVOID is a DECC funded initiative which allows fine tuning of research tools and outputs for the use of policy makers. A second web portal, CLIMASCOPE, is being developed to allow a wide range of users free access to regional climate change projections and encourage risk assessment through encapsulation of uncertainties. We have begun discussions on appropriate design and use of this tool with stakeholders located in Mexico and Madagascar. Both the AVOID and CLIMASCOPE discussions provide fora in which we can come to better understand stakeholder needs, and use this knowledge to guide the evolving design of CIAS.

Comprehensive assessments of uncertainty in climate prediction models should in principle consider contributions from the discretised numerical schemes, as well as from the parameterised physics and other sources. The numerical contributions are often assumed to be negligible, however. This talk reviews the evidence for uncertainty arising from time-stepping schemes, and suggests a possible avenue for progress to reduce it.

The context for the progress is that many climate models use the simple leapfrog scheme in concert with the Robert-Asselin filter. Unfortunately, this filter introduces artificial damping and degrades the formal accuracy, because of a conservation problem. A simple modification to the filter is proposed, which fixes the conservation problem, thereby reducing the artificial damping and increasing the formal accuracy.

The Robert-Asselin-Williams (RAW) filter may easily be incorporated into existing climate models, via the addition of only a few lines of code that are computationally very inexpensive. Results will be shown from recent implementations of the RAW filter in various models. The modification will be shown to reduce model biases and to significantly improve the predictive skill.

The necessity to provide climate information (historical and future climates) to support decision and policy making, particularly in the context of adaptation, raises many challenges. These challenges arise as there is an increasing need for information that can inform decisions and policy development and that same time reflects the evolving state of the climate science and adaptation theory and practice. These challenges include understanding uncertainties and different means of reflecting these in adaptation assessments. Addressing these challenges often means understanding the nature and processes involved in making the associated decisions and policies. It also means looking at the information from the perspective of extracting that which is necessary to inform these processes; more than just providing a description of the climate. There is also a need to for decisions and policies to ‘embrace’ the uncertainties. This presentation explores some of the lessons learned and the challenges still needing to be addressed towards achieving the desired balance and to enhance the value of climate information in supporting adaptation.

Agricultural development in many parts of the tropics, particularly sub-Saharan Africa, faces daunting challenges, which climate change and increasing climate variability will compound in vulnerable areas. A plus-two degree world appears inevitable, and planning for and implementing successful adaptation strategies is critical if agricultural growth is to occur, food security be achieved, and household livelihoods be enhanced. Reliable climate and weather information, at various spatial and temporal scales, is essential in the design of effective strategies for adaptation to climate change and to better our understanding of the critical thresholds in global and regional food systems. Land-based observation and data collection systems in parts of the tropics have been in decline for decades, and the uncertainty in basic information adds considerable difficulty to the quantification and evaluation of impacts and adaptation options. While new technology is being brought to bear on the data issue, much remains to be done. Serious attention also needs to be given to the evaluation of uncertainty and its presentation to decision-makers, in a way that can facilitate the design and implementation of policy-relevant action.

While the EU has been propagating the 2° target over the last decade, now also the Copenhagen Accord (2009) recognizes that 'the increase in global temperature should be below 2 degrees Celsius' (compared to pre-industrial levels). In recent years, energy economics have derived welfare-optimal investment streams into low-emission energy mixes and associated costs. According to our analyses, auxiliary targets that are in line with the 2° target could be achieved at relatively low costs if energy investments were triggered rather swiftly.

While such analyses assume 'perfect foresight' of a benevolent 'social planner', an accompanying suite of experiments explicitly acknowledges the rather uncertain nature of key responses to human decisions within the climate as well as the technology system. We outline first results into that direction and indicate an intrinsic need for generalisation within target approaches under uncertainty.