The impact of parametric uncertainty and topographic error in ice-sheet modelling

Hebeler, Felix; Purves, Ross S.; Jamieson, Stewart S. R.

doi:10.3189/002214308787779852

Cited by 26 publications

(30 citation statements)

References 62 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This is made more difficult when little is known about the a priori behavior as parameter ranges need to be explored evenly. Such uncertain parameters mean that expert judgments on relative parameter importance, often used to reduce the dimensionality of ensemble experiments [Hebeler et al, 2008;Murphy et al, 2004], are harder to justify. This study aims to tackle the problem by identifying relative parameter importance using a "parameter screening" experiment.…”

Section: Parameter Screeningmentioning

confidence: 99%

“…Forcing data and initialization uncertainty are often dealt with in a proceeding sensitivity analysis, so for the purpose of this method they will be fixed and only sensitivities due to the internal physical parameters (x 1 , x 2 , …, x k ) will be explored. However, as with all physically based model inputs the appropriate value is imperfectly known and is described by a range [Hebeler et al, 2008;Murphy et al, 2004;Oakley and O'Hagan, 2004]. This method presupposes that the parameters are "variation independent," which means that every point in the input uncertainty space is reasonable within the scientific literature, current expert judgment and physical applicability.…”

Section: Outlinementioning

confidence: 99%

“…A value of 0.6 is commonly used in parameterized snow models and has been extensively implemented in distributed ice sheet runs [Huybrechts et al, 1991;Robinson et al, 2010;Rutt et al, 2009]. Hebeler et al [2008] carried out a parametric uncertainty analysis of a PDD melt model and varied r max between 0.4 and 0.8. This study chose to use a range of 0.3-0.9 using a simple 50% scaling argument from the commonly used value of 0.6.…”

Section: A2 Refreezing Parameters: R Maxmentioning

confidence: 99%

See 2 more Smart Citations

Exploration of parametric uncertainty in a surface mass balance model applied to the Greenland ice sheet

et al. 2012

View full text Add to dashboard Cite

[1] Computational modeling of Earth system processes often requires simplifying assumptions of the real system. These necessary assumptions result in the definition of internal model parameters that can take a number of different values but must be explicitly defined for any one model simulation. The main issue with such an uncertain multidimensional input space is that many simulations are needed to adequately explore it. This study presents a generalized parameter screening experiment for use in future earth system modeling. This approach identifies model parameters that dominate uncertainty, therefore reducing to a manageable number the simulations required to explore the input space. The approach we adopt is relatively inexpensive to implement and can be applied at both the aggregate and disaggregate (e.g., regional) level. To demonstrate the potential of such a method, it is applied to a surface mass balance model of intermediate complexity over the Greenland ice sheet. All identified parameters were related to the surface melt parameterization, with albedo parameters being identified as the most important. Spatial distributions of the parameter sensitivities show that, in recent years, most parameter sensitivities are concentrated around the southwest and northern ice sheet margins. Simulations for the 21st century indicate an increase in sensitivity in these high melt areas especially in the northeast. Melt contributions from temperature and radiative effects are shown to be important on the order of parameters identified, and as a consequence, sensitivities are dependent on the present climate used for modeling surface mass balance.Citation: Fitzgerald, P. W., J. L. Bamber, J. K. Ridley, and J. C. Rougier (2012), Exploration of parametric uncertainty in a surface mass balance model applied to the Greenland ice sheet,

show abstract

Section: Parameter Screeningmentioning

confidence: 99%

Section: Outlinementioning

confidence: 99%

Section: A2 Refreezing Parameters: R Maxmentioning

confidence: 99%

See 1 more Smart Citation

Exploration of parametric uncertainty in a surface mass balance model applied to the Greenland ice sheet

et al. 2012

View full text Add to dashboard Cite

show abstract

“…The uncertainty of a w is estimated by propagating the uncertainty inherent in the air temperature measurements and the lapse rate. The lapse rate is assumed to be normally distributed with mean equal to the standard value of −6.5 K km −1 for the Alps and standard deviation of 1 K km −1 , based on the investigations of Hebeler and Purves (2008). Following the investigations of Foster et al (2006), a w and w are assumed to be lognormally distributed.…”

Section: Water Vapormentioning

confidence: 99%

Uncertainties of parameterized surface downward clear-sky shortwave and all-sky longwave radiation.

2012

View full text Add to dashboard Cite

Abstract.As many environmental models rely on simulating the energy balance at the Earth's surface based on parameterized radiative fluxes, knowledge of the inherent model uncertainties is important. In this study we evaluate one parameterization of clear-sky direct, diffuse and global shortwave downward radiation (SDR) and diverse parameterizations of clear-sky and all-sky longwave downward radiation (LDR). In a first step, SDR is estimated based on measured input variables and estimated atmospheric parameters for hourly time steps during the years 1996 to 2008. Model behaviour is validated using the high quality measurements of six Alpine Surface Radiation Budget (ASRB) stations in Switzerland covering different elevations, and measurements of the Swiss Alpine Climate Radiation Monitoring network (SACRaM) in Payerne. In a next step, twelve clear-sky LDR parameterizations are calibrated using the ASRB measurements. One of the best performing parameterizations is elected to estimate all-sky LDR, where cloud transmissivity is estimated using measured and modeled global SDR during daytime. In a last step, the performance of several interpolation methods is evaluated to determine the cloud transmissivity in the night.We show that clear-sky direct, diffuse and global SDR is adequately represented by the model when using measurements of the atmospheric parameters precipitable water and aerosol content at Payerne. If the atmospheric parameters are estimated and used as a fix value, the relative mean bias deviance (MBD) and the relative root mean squared deviance (RMSD) of the clear-sky global SDR scatter between between −2 and 5 %, and 7 and 13 % within the six locations. The small errors in clear-sky global SDR can be attributed to compensating effects of modeled direct and diffuse SDR since an overestimation of aerosol content in the atmosphere results in underestimating the direct, but overestimating the diffuse SDR. Calibration of LDR parameterizations to local conditions reduces MBD and RMSD strongly compared to using the published values of the parameters, resulting in relative MBD and RMSD of less than 5 % respectively 10 % for the best parameterizations. The best results to estimate cloud transmissivity during nighttime were obtained by linearly interpolating the average of the cloud transmissivity of the four hours of the preceeding afternoon and the following morning.Model uncertainty can be caused by different errors such as code implementation, errors in input data and in estimated parameters, etc. The influence of the latter (errors in input data and model parameter uncertainty) on model outputs is determined using Monte Carlo. Model uncertainty is provided as the relative standard deviation σ rel of the simulated frequency distributions of the model outputs. An optimistic estimate of the relative uncertainty σ rel resulted in 10 % for the clear-sky direct, 30 % for diffuse, 3 % for global SDR, and 3 % for the fitted all-sky LDR.

show abstract

“…It has recently been employed by van den Berg et al (2008) to simulate ice sheet changes through glacial cycles. Such a parameterization, introduced early on by Pollard (1980), found its application to the simulation of the evolution of ice sheets during the ice ages (Esch and Herterich, 1990;Deblonde and Peltier, 1992;Peltier and Marshall, 1995) and is nowadays becoming more prevalent in ice sheet modeling (e.g., Hebeler et al, 2008). However, this method is still not as widely used or accepted as the PDD method.…”

mentioning

confidence: 99%

An efficient regional energy-moisture balance model for simulation of the Greenland Ice Sheet response to climate change

2010

View full text Add to dashboard Cite

Abstract. In order to explore the response of the Greenland ice sheet (GIS) to climate change on long (centennial to multi-millennial) time scales, a regional energy-moisture balance model has been developed. This model simulates seasonal variations of temperature and precipitation over Greenland and explicitly accounts for elevation and albedo feedbacks. From these fields, the annual mean surface temperature and surface mass balance can be determined and used to force an ice sheet model. The melt component of the surface mass balance is computed here using both a positive degree day approach and a more physically-based alternative that includes insolation and albedo explicitly. As a validation of the climate model, we first simulated temperature and precipitation over Greenland for the prescribed, present-day topography. Our simulated climatology compares well to observations and does not differ significantly from that of a simple parameterization used in many previous simulations. Furthermore, the calculated surface mass balance using both melt schemes falls within the range of recent regional climate model results. For a prescribed, ice-free state, the differences in simulated climatology between the regional energy-moisture balance model and the simple parameterization become significant, with our model showing much stronger summer warming. When coupled to a threedimensional ice sheet model and initialized with present-day conditions, the two melt schemes both allow realistic simulations of the present-day GIS.

show abstract

The impact of parametric uncertainty and topographic error in ice-sheet modelling

Cited by 26 publications

References 62 publications

Exploration of parametric uncertainty in a surface mass balance model applied to the Greenland ice sheet

Exploration of parametric uncertainty in a surface mass balance model applied to the Greenland ice sheet

Uncertainties of parameterized surface downward clear-sky shortwave and all-sky longwave radiation.

An efficient regional energy-moisture balance model for simulation of the Greenland Ice Sheet response to climate change

Contact Info

Product

Resources

About