The Greenland Ice Sheet is the world's largest single source of barystatic sea-level rise (c.20% total rise) and more than half of the mass lost annually from the ice sheet comes from surface melt-water runoff. This proportion, and its magnitude, is rising with continued climate warming but future projections, and societal planning for sea level rise impacts, are undermined by a fundamental source of uncertainty. Across the vast majority of the accumulation area of the Greenland Ice Sheet, we do not know how much of the water produced from surface melting refreezes in underlying firn (i.e. multi-year snow) or becomes runoff. When the surface of an ice sheet melts, the density and temperature of underlying snow, firn and impermeable ice combine to determine whether melt refreezes in the underlying snow and firn, or becomes runoff to the ocean. If meltwater can percolate to depth (e.g. up to c.10 m) and access cold, low density firn, it can refreeze creating a significant buffer between climate change and sea-level rise. Alternatively, if melt encounters shallow impermeable ice layers (themselves created by previous refreezing) within relatively warm firn, melt cannot reach the cold firn and more melt will become runoff. The difference between these two scenarios alone could double ice sheet runoff by the middle of the 21st century. We rely on model simulations of surface melt, refreezing and runoff to accurately project the future contribution of the Greenland Ice Sheet to sea level rise. However, model-based estimates of the annual refreezing capacity of the ice sheet over the last six decades differ dramatically and undermines their ability to converge towards a reliable range of future projections. A major cause of uncertainty follows from the quite different assumptions that models make about ice layer permeability that dramatically alters the ice sheet refreezing capacity. If ice layers in firn are assumed to be impermeable (permeable), they will inhibit (allow) meltwater percolation to depth, diminish (maintain) refreezing capacity, increase (decrease) runoff and hence increase (decrease) projected global sea level rise. Without an improved treatment of ice layer permeability, existing surface mass balance models cannot provide reliable projections of the future refreezing capacity of, and melt-water runoff from, the Greenland Ice Sheet, leaving the ice sheet's future contribution to sea level rise highly uncertain. Firstly, we need to know the physical and thermal conditions of snow and firn that control the effective permeability of relatively thin ice layers (<0.5m thick) since within our warming climate these are increasingly determining the depth to which meltwater can percolate and hence control the refreezing capacity of the underlying firn. To this end we will undertake temperature-controlled laboratory experiments, systematically simulating and monitoring snow/firn/ice melt/refreezing/runoff. Secondly, we need to model the effective permeability of ice layers in snow and firn and their sensitivity to changing external and internal conditions since these together control how much melt refreezes or becomes runoff. For this, our lab work will inform novel developments to modelling to simulate measured arctic ice cap snowpack evolution. Finally we will incorporate improved ice layer permeability criteria within ice sheet scale models of the Greenland Ice Sheet to generate more accurate simulations of runoff and refreezing during melt extremes and improve harmonisation of long-term mass balance model projections, consequently improving global sea level rise predictions over the next century. Multiple recent "exceptional" melt seasons have caused near surface ice layers to proliferate through previously low density firn. These extremes will be the new norm in the future so new model parameterisations are urgently required that can effectively characterise ice layer control on mass balance.

Surface temperature is the longest instrumental record of climate change and the measure used in the Paris Climate Agreement that aims to 'prevent dangerous anthropogenic interference with the climate system'. The Agreement defines an ambition to limit global temperature change to 1.5C or 2C above pre-industrial levels. The Intergovernmental Panel on Climate Change (IPCC) used a baseline of 1850-1900 for its definition of 'pre-industrial' as this is when existing instrumental records begin. It has been estimated that global temperatures may have already increased by 0.0-0.2C by this time, but this is uncertain due to lack of data. However, even using the 1850-1900 baseline, existing temperature datasets disagree on the amount of warming to date and this disagreement implies more than 20% uncertainty in the allowed carbon budget to meet the goals of the Paris Agreement solely due to uncertainty in observed surface temperature change. These differences between temperature datasets arise mostly from two structural uncertainties: the use of sea surface temperatures (SST) rather than air temperatures over the oceans, especially ice-covered regions, and differences in data coverage and interpolation strategies. This project addresses both. To best inform decision-makers, records of temperature change must be as accurate, consistent, and long as possible. Existing global datasets start in 1850 or later, but we will extend the record a further 70 years back to the late 18th century. Current knowledge of this period comes from instrumental measurements in Europe, palaeo-proxies (tree-rings, corals or ice cores), and climate models. We will dramatically extend the spatial coverage of the early measured record in this 70-year period, which is important for understanding natural climate variability and the climate response to different radiative forcings. For example, the longer record includes the period of 5 large volcanic eruptions and extra cycles of multi-decadal climate oscillations. The new record will allow us to better disentangle the contributions of anthropogenic and natural factors on the climate system and quantify the effect humans have already had on Earth's temperature, and hence on future climate. A major inconsistency has been past use of air temperature over land but SST over oceans. Recent advances mean we can produce a marine air temperature record to construct the first global air temperature dataset over ocean, land and ice, stretching back to the late 18th century. Our dataset will be independent from SST, currently the most uncertain component of global temperature. We will improve land, marine and cryosphere air temperature observations to make them more homogeneous and extend the global record further back in time. This requires fundamental research to better understand the bias and noise characteristics of historical observations and develop new error models. We will adopt sophisticated statistical techniques to allow the estimation of air temperature everywhere, even when there are gaps in the observations. We will expand the historical climate record with new ship's logbook and weather station digitisations focused on early data, sparse periods and regions, and the interfaces between land, ocean and ice. We will engage the public in the digitisation effort building on recent successful citizen science initiatives. We will analyse the new surface air temperature record to better understand how temperatures have changed since the late 18th century. This longer record will give a better understanding of natural climate variations, both variability generated internally within the climate system and that due to external forcing factors such as volcanic eruptions and solar changes. This improved understanding of natural variability will enable us to more cleanly isolate the characteristic "fingerprints" of man-made climate change allowing us to more confidently detect and attribute human-induced changes

Surface temperature is the longest instrumental record of climate change and the measure used in the Paris Climate Agreement that aims to 'prevent dangerous anthropogenic interference with the climate system'. The Agreement defines an ambition to limit global temperature change to 1.5C or 2C above pre-industrial levels. The Intergovernmental Panel on Climate Change (IPCC) used a baseline of 1850-1900 for its definition of 'pre-industrial' as this is when existing instrumental records begin. It has been estimated that global temperatures may have already increased by 0.0-0.2C by this time, but this is uncertain due to lack of data. However, even using the 1850-1900 baseline, existing temperature datasets disagree on the amount of warming to date and this disagreement implies more than 20% uncertainty in the allowed carbon budget to meet the goals of the Paris Agreement solely due to uncertainty in observed surface temperature change. These differences between temperature datasets arise mostly from two structural uncertainties: the use of sea surface temperatures (SST) rather than air temperatures over the oceans, especially ice-covered regions, and differences in data coverage and interpolation strategies. This project addresses both. To best inform decision-makers, records of temperature change must be as accurate, consistent, and long as possible. Existing global datasets start in 1850 or later, but we will extend the record a further 70 years back to the late 18th century. Current knowledge of this period comes from instrumental measurements in Europe, palaeo-proxies (tree-rings, corals or ice cores), and climate models. We will dramatically extend the spatial coverage of the early measured record in this 70-year period, which is important for understanding natural climate variability and the climate response to different radiative forcings. For example, the longer record includes the period of 5 large volcanic eruptions and extra cycles of multi-decadal climate oscillations. The new record will allow us to better disentangle the contributions of anthropogenic and natural factors on the climate system and quantify the effect humans have already had on Earth's temperature, and hence on future climate. A major inconsistency has been past use of air temperature over land but SST over oceans. Recent advances mean we can produce a marine air temperature record to construct the first global air temperature dataset over ocean, land and ice, stretching back to the late 18th century. Our dataset will be independent from SST, currently the most uncertain component of global temperature. We will improve land, marine and cryosphere air temperature observations to make them more homogeneous and extend the global record further back in time. This requires fundamental research to better understand the bias and noise characteristics of historical observations and develop new error models. We will adopt sophisticated statistical techniques to allow the estimation of air temperature everywhere, even when there are gaps in the observations. We will expand the historical climate record with new ship's logbook and weather station digitisations focused on early data, sparse periods and regions, and the interfaces between land, ocean and ice. We will engage the public in the digitisation effort building on recent successful citizen science initiatives. We will analyse the new surface air temperature record to better understand how temperatures have changed since the late 18th century. This longer record will give a better understanding of natural climate variations, both variability generated internally within the climate system and that due to external forcing factors such as volcanic eruptions and solar changes. This improved understanding of natural variability will enable us to more cleanly isolate the characteristic "fingerprints" of man-made climate change allowing us to more confidently detect and attribute human-induced changes

Sea levels will rise significantly in the coming decades as a result of greenhouse gas emissions, but there are large uncertainties about how fast they will rise in different emission futures. In a warming climate, the main causes of sea level rise are thermal expansion of sea water, melting of glaciers and ice sheets, and ice-flow directly into the oceans (dynamic ice loss). Dynamic ice loss from the Greenland and Antarctic Ice Sheets is, by a large margin, the greatest source of uncertainty in predictions of sea level rise, and ranges from 10 cm to over 1 m by 2100 in a high emissions scenario. The upper and lower limits of this range have very different implications for coastal communities and economies, hampering efforts to plan for the future. Identifying safe limits on carbon emissions and adopting appropriate mitigation strategies require reliable predictions of dynamic ice loss from the ice sheets. Dynamic ice loss is complex because it is controlled by the fracture and detachment of icebergs (calving) and the submarine melt of ice in contact with the ocean. Calving and melting can reduce resistance to ice flow, leading to faster transfer of mass from land to the oceans and potentially to irreversible ice-sheet collapse. Despite their importance, calving and submarine melting are very poorly represented in the models used to predict ice-sheet response to climate change, resulting in high uncertainties in future dynamic ice loss and hence sea-level rise. There is an urgent need to develop reliable calving laws for ice sheet-models, based on a thorough understanding of calving processes and their interactions with ice flow and submarine melt. We aim to solve this problem using a new high-resolution model of fracturing, ice-dynamics and ocean processes (FIDO), to build a solid foundation for the development of the calving laws required for predictive ice-sheet models. FIDO combines state-of-the-art methods of modelling ice-flow and ocean circulation with a revolutionary ice-fracture model. Unlike conventional approaches, the fracture model represents ice as assemblages of particles linked by breakable bonds - much like real ice - allowing calving to be simulated with impressive realism. We will use FIDO to simulate calving, submarine melt and ice flow in a wide range of environmental conditions, and rigorously test the validity of the results using satellite observations of ice margin behaviour in Greenland and Antarctica. From the FIDO model results, we will distil the essential rules of calving and define new, comprehensive calving laws incorporating interactions with submarine melt and ice dynamics. In collaboration with UK and international partners, we will implement our new calving laws in models of the Greenland and Antarctic Ice Sheets, to predict their 21st Century sea-level contributions for a range of greenhouse gas emission scenarios. We anticipate radically improved sea-level rise predictions by 2020, in readiness for the 6th Assessment Report of the Intergovernmental Panel on Climate Change (IPCC AR6).

The atmosphere changes on time scales from seconds (or less) through to years. An example of the former are leaves swirling about the ground within a dust-devil, while an example of the latter is the quasibiennial oscillation (QBO) which occurs over the equator high up in the stratosphere. The QBO is seen as a slow meander of winds: from easterly to westerly to easterly over a time scale of about 2.5 years. This 'oscillation' is quite regular and so therefore is predictable out from months through to years. These winds have also been linked with weather events in the high latitude stratosphere during winter, and also with weather regimes in the North Atlantic and Europe. It is this combination of potential predictability and the association with weather which can affect people, businesses and ultimately economies which makes knowing more about these stratospheric winds desirable. However, it has been difficult to get this phenomenon reproduced in global climate models. We know that to get these winds in models one needs a good deal of (vertical) resolution. Perhaps better than 600-800m vertical resolution is needed. In most GCMs with a QBO this is the case, but why? We also know that there needs to be waves sloshing about, either ones that can be 'seen' in the models, or wave effects which are inferred by parameterisations. Get the right mix of waves and you can get a QBO. Get the wrong mix and you don't. Again we do not know entirely why. Furthermore, we also know convection bubbling up over the tropics and the slow migration of air upwards and out to the poles also has a big impact of resolving the QBO. All of these factors need to be constrained in some way to get a QBO. The trouble is that these factors are invariably different in different climate models. It is for this reason that getting a regular QBO in a climate model is so hard. This project is interested in exploring the sensitivity of the QBO to changes in resolution, diffusion and physics processes in lots of climate models and in reanalyses (models used with observations). To achieve this, we are seeking to bring together all the main modelling centres around the world and all the main researchers interested in the QBO to explore more robust ways of modelling this phenomena and looking for commonalities and differences in reanalyses. We hope that by doing this, we may get more modelling centres interested and thereby improve the number of models which can reproduce the QBO. We also hope that we can get a better understanding of those impacts seen in the North-Atlantic and around Europe and these may affect our seasonal predictions. The primary objective of QBOnet is to facilitate major advances in our understanding and modelling of the QBO by galvanizing international collaboration amongst researchers that are actively working on the QBO. Secondary objectives include: (1) Establish the methods and experiments required to most efficiently compare dominant processes involved in maintaining the QBO in different models and how they are modified by resolution, numerical representation and physics parameterisation. (2) Facilitate (1) by way of targeted visits by the PI and researchers with project partners and through a 3-4 day Workshop (3) Setup and promote a shared computing resource for both the QBOi and S-RIP QBO projects on the JASMIN facility