Periodic Reporting for period 4 - PALGLAC (Palaeoglaciological advances to understand Earth’s ice sheets by landform analysis)
Berichtszeitraum: 2023-04-01 bis 2024-09-30
Ice sheets regulate Earth’s climate by reflecting sunlight away, enabling suitable temperatures for human habitation. Warming is reducing these ice masses and raising sea level. Glaciologists predict ice loss using computational ice sheet models which interact with climate and oceans, but with caveats highlighting that some processes are inadequately encapsulated. Weather forecasting made a leap in skill by comparing modelled forecasts with actual outcomes to improve physical realism of their models. This project sets out to adopt this data-modelling approach in ice sheet modelling. Given their longer timescales (100-1000s years) we will use geological and geomorphological records of former ice sheets to provide the evidence; the rapidly growing field of palaeoglaciology.
Focussing on the most numerous and spatially-extensive records of palaeo ice sheet activity - glacial landforms - the project aims to revolutionise understanding of past, present and future ice sheets. Our mapping campaign (Work-Package 1), complemented by machine learning techniques (WP2), will vastly increase the evidence-base. Resolution of how subglacial landforms are generated and how hydrological networks develop (WP3) would be major breakthroughs with key implications for ice flow models and hydrological effects on ice dynamics. By pioneering techniques and coding for combining ice sheet models with landform data (WP4) we will improve knowledge of the role of palaeo-ice sheets in Earth system change. Trialling of numerical models in these data-rich environments will highlight deficiencies in process-formulations, leading to better models.
Focussing on the most numerous and spatially-extensive records of palaeo ice sheet activity - glacial landforms - the project aims to revolutionise understanding of past, present and future ice sheets. Our mapping campaign (Work-Package 1), complemented by machine learning techniques (WP2), will vastly increase the evidence-base. Resolution of how subglacial landforms are generated and how hydrological networks develop (WP3) would be major breakthroughs with key implications for ice flow models and hydrological effects on ice dynamics. By pioneering techniques and coding for combining ice sheet models with landform data (WP4) we will improve knowledge of the role of palaeo-ice sheets in Earth system change. Trialling of numerical models in these data-rich environments will highlight deficiencies in process-formulations, leading to better models.
Many tens of terabytes of metre-resolution digital elevation data were investigated to find and map over half a million glacial landforms including those recording ice flow direction (such as drumlins), subglacial water flow (eskers and meltwater corridors) and successive ice margin positions (end moraines). This was achieved for the whole of Norway, Sweden, and Finland and glaciated parts of the Netherlands, Denmark, Germany, Poland, Lithuania, Latvia, Estonia, and Russia.
Using our observations and computational modelling we have advanced understanding of how these landforms are created, notably for drumlins and mega scale glacial lineations and for landforms arising from water drainage beneath ice sheets. We have built reconstructions combining landform data, geochronological data on timing (e.g. by radiocarbon dating) and computational ice sheet modelling for the last British-Irish Ice Sheet and for the ice sheet that covered Greenland from the last glacial, and have nearly completed this task for the former Scandinavian Ice Sheet.
For the former British-Irish Ice Sheet of the last glaciation (31 000 to 15 000 years ago) information on timing of retreat gathered on a prior project (BRITICE-CHRONO) was used on PalGlac with our tools for combining data and modelling to yield reconstructions of the changing ice extent, thickness and flow velocity through time. This is the first time a model has been tested against so much empirical data including flow direction, ice extent and timing, and this model represents the most well-constrained reconstruction of a retreating ice sheet anywhere in the world. We also modelled how the earth’s crust and its topography varied as a consequence of mass loading by this ice sheet. These sea level aspects are being used by those interested in forecasting future sea levels, relevant for low lying parts of Europe.
Our mapping of 190,000 glacial landforms recording former ice margin positions (moraines etc.) around the present day Greenland Ice Sheet was used to address an important hypothesis relevant for forecasting ice loss to year 2100, and consequent sea level rise. By combining the landforms with published dates on the timing of retreat we built maps of the retreating ice sheet at 500 year time-steps from 16,000 years ago. In order to ask whether such palaeo-history matters for future projections of mass loss from the Greenland Ice Sheet, we compared simulations accounting for this history against a simulation with no history and starting from the modern ice sheet in steady state. We found large differences, with the simulation using palaeo-history predicting the magnitude of sea level rise twice that of the simulation without prior history. This is a very important finding because it shows that more robust sea level forecasts require inclusion of prior ice sheet history over 1000s of years, and underlines the significance of improving palaeoglaciological knowledge of Earth’s ice sheets; the central aim of the PalGlac project. We conducted two field expeditions to Greenland to collect rock samples for cosmogenic dating, supporting our work to better constrain the timing of retreat of this ice sheet.
As the Scandinavian Ice Sheet grew and decayed, its patterns of ice flow and flow divides shifted and migrated. We argue that such geometrical properties are more important to capture in ice sheet modelling than just ice margin positions, which by their very nature are difficult to model; where ice tends to 0 m thickness. We pioneered a new approach to combine and steer model simulations using empirical observations of ice flow compiled from landforms such as drumlins. This work required making new comparison tools that are more statistically robust than previously adopted and devising a novel method of incorporating (calibrating) observed ice flow directions within an ice sheet model. These approaches are currently being used to build a flow-optimised reconstruction of the Scandinavian Ice Sheet that also seeks to match the geochronological data on ice margin timing.
Using our observations and computational modelling we have advanced understanding of how these landforms are created, notably for drumlins and mega scale glacial lineations and for landforms arising from water drainage beneath ice sheets. We have built reconstructions combining landform data, geochronological data on timing (e.g. by radiocarbon dating) and computational ice sheet modelling for the last British-Irish Ice Sheet and for the ice sheet that covered Greenland from the last glacial, and have nearly completed this task for the former Scandinavian Ice Sheet.
For the former British-Irish Ice Sheet of the last glaciation (31 000 to 15 000 years ago) information on timing of retreat gathered on a prior project (BRITICE-CHRONO) was used on PalGlac with our tools for combining data and modelling to yield reconstructions of the changing ice extent, thickness and flow velocity through time. This is the first time a model has been tested against so much empirical data including flow direction, ice extent and timing, and this model represents the most well-constrained reconstruction of a retreating ice sheet anywhere in the world. We also modelled how the earth’s crust and its topography varied as a consequence of mass loading by this ice sheet. These sea level aspects are being used by those interested in forecasting future sea levels, relevant for low lying parts of Europe.
Our mapping of 190,000 glacial landforms recording former ice margin positions (moraines etc.) around the present day Greenland Ice Sheet was used to address an important hypothesis relevant for forecasting ice loss to year 2100, and consequent sea level rise. By combining the landforms with published dates on the timing of retreat we built maps of the retreating ice sheet at 500 year time-steps from 16,000 years ago. In order to ask whether such palaeo-history matters for future projections of mass loss from the Greenland Ice Sheet, we compared simulations accounting for this history against a simulation with no history and starting from the modern ice sheet in steady state. We found large differences, with the simulation using palaeo-history predicting the magnitude of sea level rise twice that of the simulation without prior history. This is a very important finding because it shows that more robust sea level forecasts require inclusion of prior ice sheet history over 1000s of years, and underlines the significance of improving palaeoglaciological knowledge of Earth’s ice sheets; the central aim of the PalGlac project. We conducted two field expeditions to Greenland to collect rock samples for cosmogenic dating, supporting our work to better constrain the timing of retreat of this ice sheet.
As the Scandinavian Ice Sheet grew and decayed, its patterns of ice flow and flow divides shifted and migrated. We argue that such geometrical properties are more important to capture in ice sheet modelling than just ice margin positions, which by their very nature are difficult to model; where ice tends to 0 m thickness. We pioneered a new approach to combine and steer model simulations using empirical observations of ice flow compiled from landforms such as drumlins. This work required making new comparison tools that are more statistically robust than previously adopted and devising a novel method of incorporating (calibrating) observed ice flow directions within an ice sheet model. These approaches are currently being used to build a flow-optimised reconstruction of the Scandinavian Ice Sheet that also seeks to match the geochronological data on ice margin timing.
We have established methods, data processing and coding that permit large volumes of landform data to be acquired, interpreted and used for scoring or guiding/constraining numerical ice sheet model simulations. This takes us beyond the state of the art, both in terms of data volume and spatial extent of the information, and by integrating previously disparate observational and modelling approaches. We have published on these methodological advances, improved statistical robustness, and PalGlac is now leading by example. Hopefully we can further direct the field down this route with large gains to be made in both simulations of Earth history and climate and for improvements in robustness of numerical modelling.
Using our landform maps and new tools for data-modelling integration we built ice sheet reconstructions for the British-Irish, Greenland and Scandinavian ice sheets. Each of these are now the state of the art for the field and I expect will serve as important benchmarks over the next few decades. They are relevant and hopefully useful for investigations wider than glaciology such as for modern and future sea level, groundwater flow, archaeology, migrations of biota, mineral exploration and substrate geology. Our finding that the thousands of years prior glacial history of the Greenland Ice Sheet really matters and significantly affects the future rate of ice loss is especially important. We hope this sets a new agenda in ice sheet modelling more closely attuned to palaeoglaciology and which leads to improvements in the robustness of sea level forecasts, of great relevance to how societies adapt to climate warming.
Using our landform maps and new tools for data-modelling integration we built ice sheet reconstructions for the British-Irish, Greenland and Scandinavian ice sheets. Each of these are now the state of the art for the field and I expect will serve as important benchmarks over the next few decades. They are relevant and hopefully useful for investigations wider than glaciology such as for modern and future sea level, groundwater flow, archaeology, migrations of biota, mineral exploration and substrate geology. Our finding that the thousands of years prior glacial history of the Greenland Ice Sheet really matters and significantly affects the future rate of ice loss is especially important. We hope this sets a new agenda in ice sheet modelling more closely attuned to palaeoglaciology and which leads to improvements in the robustness of sea level forecasts, of great relevance to how societies adapt to climate warming.