Overview of the Ice Mass Balance Inter-comparison Exercise (IMBIE)

    project findings

--- Erik Ivins, JPL, Caltech, USA, and Ben Smith, APL, University of Washington, USA)

Abstract

The Ice-sheet Mass Balance Inter-comparison Exercise (IMBIE) is a combined ESA/NASA effort to perform experiments that would allow better understanding of the discrepant results that have arisen concerning the mass balance of the Greenland (GIS) and Antarctic (AIS) ice–sheets, including quite different quantitative assessments of east (EAIS) and west (WAIS) for the latter. Three different techniques are inter-compared, with intra-comparison among different groups using the same space data. The techniques are ‘Input-Output Method’ (IOM), radar and laser altimetry (RA and LA), and GRACE gravimetry, and the periods open for experimentation are 1992-2011, with special focus on the period 2003-2008, when direct intercomparison among techniques is feasible.


A series of experiments established grounds for the inter-comparison. Direct comparisons between RA and IOM for basins within the AIS showed that the two techniques give largely compatible results, with an average difference between the two of 1.4 ± 3.8 Gt /yr and agreement to within the 2-σ errors for 42 of 49 basins. Rates for regions surveyed by RA and not by IOM were small (4.5 ± 6.0 for WAIS and 1.4 ±1.7 Gt /yr for EAIS) demonstrating that direct IOM measurements capture nearly all significant mass loss in Antarctica. Likewise, comparisons between modeled SMB (as used for IOM), RA, and GRACE measurements during an exceptional 2009 EAIS snowfall event showed close agreement between the magnitude and spatial pattern of the resulting mass anomaly.


The different techniques then were used to establish mass-rate estimates suitable for comparison and for the establishment of reconciled estimates. IMBIE IOM results include updated error estimates, but exclude heuristic scaling between measured and unmeasured areas that have been used in other publications, giving somewhat less negative results than have been reported elsewhere. RA gave estimates of WAIS and EAIS elevation rates north of 86 S. For regions where ice dynamics suggest changes in discharge, the volume estimate was scaled by the density of ice. Where as, elsewhere it was scaled by the density of snow to produce masschange estimates. Laser altimetry data were analyzed by four separate groups using different techniques to interpolate between measurements, and two sets of techniques to correct for instrumental biases and to scale between volume and mass changes; these results were limited to 2003-08, but covered as far north and south as 86 degrees. Different LA groups found significantly different mean rates of mass change, chiefly in East Antarctica where correction magnitudes are large compared to elevation-change signals. An additional area of debate is over the role of the glacial isostatic adjustment (GIA), as it has an important influence on GRACE results reported for Antarctica. Largely due to maturation of two ancillary data sets (Whitehouse – this session), two new GIA (nGIA) models can be folded into our analyses to replace two older models. Use of nGIA improves our understanding of GRACE’s constraint on AIS. Combining mass-change estimates for the 2003-08 period showed all three techniques agreeing for all of the regions to within estimated errors, except for LA on the EAIS where corrections tended to dominate the signal. Combined AIS and GIS IMBIE time series give an accelerating rate of loss, with mass rates of -102 ± 118 between 1992 and 2000, -292 ± 75 between 2000 and 2012, and -343 ± 62 Gt /yr between 2005 and 2012.


The optimum target period for GRACE intra-comparison among six research groups was 2003- 2010. The results, with nGIA models used for Antarctica, gave WAIS = -108 ± 26, EAIS = +57 ± 35 and AIS = -81 ± 35 Gt /yr, while for Greenland the loss determined in the experiment is much larger: GIS = -229 ± 27 Gt /yr. Clear acceleration of loss in WAIS and GIS, and a gain in EAIS, over the target time period is identified for the intra-comparison.