cc: jalowe@meto.gov.uk, gjjenkins@meto.gov.uk, s.raper@uea.ac.uk date: Thu, 17 Jun 1999 13:04:22 +0100 (BST) from: Jonathan Gregory subject: S550 and S750 data to: m.hulme@uea.ac.uk, Rjnicho@aol.com Dear Mike and Robert In the following emails I am sending four text files, for S550 and S750 1990- 2250 calculated by Jason and me using two different ice melt parametrisations from HadCM2 surface air temperature changes. Before 1990 both scenarios are identical with GHG. (1) Oerlemans-IPCC. The glacier and Greenland scheme of Gregory and Oerlemans (1998), with the Antarctic mass balance sensitivity from the IPCC SAR. (2) Wigley-Raper. The schemes used by the Wigley-Raper model of the SAR, with parameters as used in the SAR and detailed in the IPCC simple models TP, in particular a total glacier volume of 0.3 m sea-level equivalent. The expansion is calculated from HadCM2 and is the same for both methods. The main difference is in the glacier results, which is why we did these two sets. The Oerlemans static-sensitivity approach neglects the loss of glacier area and volume as time passes, so melting continues steadily throughout the run. The Wigley-Raper model has characteristic timescales for loss of glacier volume. By 2200 practically all glacier volume has been lost. Hence, the glacier prediction of this model at 2250 is equal to its initial value of 0.3 m. This number is not well known, but in the SAR and TAR the preferred value is 0.5 m, which is quite near (though with a different time-development) to the value from the Oerlemans scheme. The Oerlemans-IPCC schemes show nearly cancelling contributions from the ice- sheets, as found by GCM experiments for the 21st century. The Wigley-Raper model has a larger +ve contribution from Greenland and a smaller -ve one from Antarctica, so they do not cancel. I would suggest that you could take the two timeseries for each scenario as alternatives, indicating systematic uncertainty in ice melt parametrisation. Best wishes Jonathan