Group picture
A group picture will come soon …
Background
In recent years, sea level rise induced by global warming and its impacts on coastal zones has become a major scientific issue involving a wide range of disciplines (IPCC, 2013) with broad societal impacts. It is now well established that global warming is essentially due to the increasing atmospheric concentration of GHG and particularly carbon dioxide from anthropogenic fossil fuel combustion and land use changes (IPCC, 2013). Global warming has already several visible consequences such as the increase of the Earth’s mean temperature and ocean heat content (more than 90% of the GHG-heat ends up in the ocean), the melting of sea ice and glaciers, the loss of ice mass from the Greenland and Antarctica ice sheets. As the oceans warm, seawater expands and sea level rises. Similarly, water from melting land ice ultimately reaches the oceans and also causes sea level to rise. Over the last 20 years, land ice melt and ocean warming caused sea level to rise at a rate over 3 mm/yr, which is an order of magnitude faster than the sea level variations over the last 3000 years, which did not exceed 0.3 mm/yr (Kemp et al. 2011). Over the 20 th century, direct sea level observations available from in situ tide gauges indicate that sea level rose actually at the rate of 1.7 +/- 0.3 mm/yr (Church and White, 2011). Since 1993, precise satellite altimeters show that sea level is rising evenfaster at a rate of 3.2 +/- 0.4 mm/yr (Meyssignac and Cazenave 2012). In addition, satellite altimetry revealed that the rate of sea level rise is not uniform over the ocean but displays large regional variations (some places experience a sea level rise 4 times faster than the global mean since 1993). All state-of-the-art climate models (i.e. GCMs) indicate that sea level will continue to rise – and more likely will accelerate – during the next decades and centuries in response to increasing GHG concentrations in the atmosphere. This future sea level rise is expected to have severe impacts on low lying islands and coastal areas including permanent submergence, increased flooding due to higher sea levels during storm surges, saltwater intrusion into surface waters and aquifers, accelerated erosion, etc
To assess the potential impacts of future sea level rise it is essential to estimate the most
accurate and reliable sea level rise projections at both global and regional scales. But current sea level rise projections are highly uncertain (Slangen et al. 2014). For any given warming scenario of the 21 st century, GCMs indicate global and regional sea level rise over the period 2081-2100 that can differ by a factor of 2 or more (IPCC, 2013). This large uncertainty arises essentially from inappropriate (or sometimes missing) model representations of some physical or dynamical processes that affect sea level in the climate system.
In a recent study, Church et al. (2013) showed that the range of an ensemble of 13 GCM
simulations of the 20 th century sea level rise encompasses the observed sea level rise within its large uncertainty. Yet this is not sufficient to make accurate, useful (in terms of adaptation and mitigation strategies) projections. Here, we want to go further and reduce the uncertainty of the ensemble of available climate models by validating each climate model estimates of the 20 th century sea level rise and its contributors against observations. The proposal consists of three phases. In the first phase (Task 1) we estimate from GCM outputs the 20 th century sea-level rise and its contributors, which are essentially ocean warming, glaciers ice melt, land water and ice-sheet mass loss. In the second and third phases (Tasks 2 and 3) we propose to compare these estimates with observations of sea level (from satellite altimetry, tide gauge records, 2D sea level reconstructions and ocean reanalysis), ocean warming (from ocean temperature records and ocean reanalysis) and land ice loss (from space gravimetry, satellite altimetry, in situ observations of glacier and ice sheet mass balance, ice sheet and glacier mass balance models forced with atmospheric reanalysis) that are available over the 20 th century.
Our approach to this problem is through an ISSI international Team since satellite-derived observations are key to all observational aspects: satellite altimetry observations of sea level changes and ice sheet surface height, space gravimetry observations of ocean, ice sheet and glaciers mass, ocean and atmospheric reanalysis products and 2D sea level reconstructions. In particular whenever available and possible, we will use the ESA-CCI essential climate variable products computed by the Climate Change Initiative project from ESA (CCI-sea level, CCI-glaciers and CCI-ice sheets) in our validation of the GCMs.
Objectives
The objective of this ISSI proposal is to (1) evaluate state-of-the-art GCMs against observations of sea level rise and (2) select those models, which best simulate the observed sea level changes and its contributors over the last 100 years. Our goal is to assess both the regional and global sea level rise in historical runs of GCMs (1860 through 2005/2014) by carefully comparing models with observations with a particular focus on near-global satellite observations.
Task 1: Sea level estimates from GCMs
Over the 20 th century the major contributors to sea level variations are 1) the thermal expansion of the ocean, 2) the glaciers mass loss, 3) the Greenland ice sheet Surface Mass Balance changes (SMB), 4) the land water storage 5) and a vertical land motion contribution from the solid Earth’s visco-elastic response to the last deglaciation and the 20 th century land ice mass loss (Gregory et al. 2013a). Antarctic ice mass loss is relatively small over the 20 th century (Huybrechts et al. 2011) and the dynamical contribution from the Greenland ice sheet is considered negligible since the Greenland ice sheet was close to equilibrium until the beginning of the 1990s (Rignot et al. 2008). Task 1 is dedicated to the computation of the 5 major contributors to the 20 th century sea level rise from outputs of GCM historical runs. These contributions will be summed up in Task 3 to estimate the sea level from each of the GCM since 1900. All CMIP5 GCMs that provide all necessary outputs will be analyzed. The methods we will apply is described in the following:
1) The global and regional thermal expansion of the ocean will be computed with the 3D ocean temperature data fields from GCM historical runs. It will be obtained by integrating the density equation from the bottom of the ocean up to the surface at each grid point. The resulting thermosteric sea level will be drift-corrected, and corrected for the cold bias due to the fact that many GCM historical runs where initialized with control runs in which volcanic forcing is omitted (see Gregory et al., 2013b for details).
2) Mountain glaciers are typically too small to be explicitly represented in the coarse CMIP5 GCMs. Thus, glacier mass loss and its contribution to sea level cannot be directly estimated from CMIP5 GCM outputs. Instead, we will use offline glacier models forced by the regional Surface Air Temperature (SAT) and precipitation outputs from CMIP5 GCMs (e.g. Marzeion et al. 2012). We propose here to test 2 different glacier models from Marzeion et al. (2012) and from Slangen and van de Wal (2011).
3) Similar to the glaciers, the Greenland SMB contribution to sea level is not correctly modeled in CMIP5 GCMs because they have a coarse horizontal spatial resolution (~300 km) which limits their capability of capturing essential SMB changes on the narrow ablation zone, and because they lack (in general) a realistic representation of the snow/firn/ice processes in the upper ice sheet. To overcome this problem, we will use different downscaling methods to estimate the Greenland SMB from the GCMs outputs. We propose to test 2 different methods: i) the downscaling method developed by Geyer et al. (in press) which gives the Greenland SMB from the SAT, the precipitation and the snowmelt outputs of the CMIP5 GCMs and ii) a regional downscaling deduced from Fettweis et al. (2013) (calibrated against the Greenland regional climate model MAR) which gives the Greenland SMB for each draining basin based on GCM precipitation and 600hPa summer temperature outputs.
4) Concerning the land-water storage contribution, it is actually mainly of anthropogenic origin over the 20 th century due to dam building and groundwater pumping (Wada et al., 2012, Konikow, 2011). This anthropogenic contribution is not modeled in CMIP5 GCMs. Thus, we will use estimates of the land water storage contribution based on 20 th century observations (from Konikow, 2011 and Wada et al., 2012) to complete our estimates of the 20 th century sea level rise from CMIP5 GCMs before comparing it to sea level observations. From 2003 onwards, we will use GRACE observations to estimate the land water component of sea level changes (e.g., Boening et al., 2012).
5) The contribution of the solid Earth’s response to the 20 th century sea level changes is probably small and close to the response of the solid Earth to the last deglaciation (20,000 years ago) only, because not much land ice has melted since 1900 (Marzeion et al. 2012). However, we will estimate it with a state-of-the-art “sea level equation” model, which takes into account the 3D deformations of the Earth, the gravitational interactions between the solid Earth, the ice sheets and the ocean and the rotation of the Earth. Giorgio Spada from the University of Urbino (Italy) has kindly agreed to provide this estimate.
The approach outlined above to compute the contributors to sea level from GCM outputs may produce different results depending on which model is used. Controlled comparisons will shed new light on the commonalities and differences among the methods and the models, will reveal robust features of the results and will help to quantify their uncertainties. These are the goals of the first task and will be addressed during the first meeting of the team.
Task 2: comparison of the sea level contributors estimates with satellite & in-situ
observations
In Task 2 we will compare estimates of the contributors to the 20 th sea level rise from CMIP5 GCMs with observations. The objective is to evaluate the GCMs performances in terms of sea level variations simulation over the 20 th century (e.g., Landerer et al., 2013). We also expect to quantify more accurately the GCMs uncertainty, to identify potential problems in GCMs simulation of sea level change and its contributors, and to point the way for both model and observations improvements as well as directions for future research and observations. Our approach is as follows:
1) The thermal expansion down to 700m depth computed from bias-corrected GCM runs will be compared with ocean temperature data from Domingues et al., 2008, Ishii and Kimoto, 2009 and Levitus et al., 2012 for the period 1961-2005 when temperature data have a nearly global coverage. Unfortunately, for the deeper ocean below 700 m, data is much more sparse and near-global coverage is only available with the Argo project since 2004. While the Argo record is short, it will still help in validating GCMs for the last decade. To validate the deeper ocean on a longer period, we propose to use 2 ocean reanalyses which assimilate temperature and salinity profiles over the period 1958-2010 plus satellite altimetry: GECCO2 (Köhl 2014) and ORAS4 (Balmaseda et al. 2013). These reanalysis data sets provide estimates of the thermal
expansion of the ocean down to the bottom and can be compared with the GCMs estimates over the last 50 years for validation purposes.
2) Over the 20 th century several sources of observations give information on glacier mass loss. These observations use different techniques and cover different time periods. In-situ observations of glacier length and altitude or glaciers models forced with atmospheric reanalysis (Marzeion et al., 2012, Cogley, 2009, Leclercq et al., 2011, Radić et Hock 2011) are two techniques, which give information over the whole 20 th century. Satellite gravimetry observations also provide some information but only since 2004 and the launch of GRACE. We intend to use these 3 techniques to validate the GCMs derived estimates of glaciers mass loss at decadal to centennial time scales. In addition, if available, we will also test the GCMs’ estimates of glaciers mass loss against the CCI-glaciers dataset.
3) Concerning the Greenland surface mass balance, direct observations are available from space with satellite altimetry since the beginning of the 1990s (CCI-ice sheets for example) and with GRACE since mid-2002. In addition, to validate the GCMs estimates over longer time periods, we will also compare the GCMs based Greenland SMB estimates with those of the regional climate model MAR (Fettweis 2013) over Greenland forced by atmospheric reanalysis over the 20 th century (from ECMWF and from NOAA). With these comparisons we expect to identify those climate models that have the smallest biases over the Greenland region and are thus best suitable for sea level estimates.
These are the goals of the second task and will be addressed during the second team meeting.
Task 3: comparison of the sea level estimates with observations
In this last task we will merge all estimates of the sea level contributors from GCMs to compute the total sea level changes for each GCM over the 20 th century and into the first decade of the 21 st century. We will then compare these estimates with observations of the sea level from tide gauge records over the 20 th century, from 2D sea level reconstructions (Church and White 2011, Hamlington et al. 2011, Meyssignac et al. 2012) over the last 60 years and from satellite altimetry over the last 20 years (with the CCI-sea level dataset). Compared to other sources of observations discussed above, sea level observations over the 20 th century are more accurate and consistent over long periods. So we expect that they will provide a more stringent validation of the GCMs and that they will help in better characterizing the GCM uncertainties. In particular, this should lead to a rigorous selection of only those GCMs which simulate climate changes over the 20 th century that are consistent with the observed sea-level rise. The outcomes of this study will then form the foundation to make revised sea level projection for 2081-2100 with refined estimates of the uncertainties and underlying causes.
These are the goals of the third task and will be addressed during the third team meeting.